emnlp emnlp2012 emnlp2012-131 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Zhongguo Li ; Guodong Zhou
Abstract: Most previous approaches to syntactic parsing of Chinese rely on a preprocessing step of word segmentation, thereby assuming there was a clearly defined boundary between morphology and syntax in Chinese. We show how this assumption can fail badly, leading to many out-of-vocabulary words and incompatible annotations. Hence in practice the strict separation of morphology and syntax in the Chinese language proves to be untenable. We present a unified dependency parsing approach for Chinese which takes unsegmented sentences as input and outputs both morphological and syntactic structures with a single model and algorithm. By removing the intermediate word segmentation, the unified parser no longer needs separate notions for words and phrases. Evaluation proves the effectiveness of the unified model and algorithm in parsing structures of words, phrases and sen- tences simultaneously. 1
Reference: text
sentIndex sentText sentNum sentScore
1 cn z Abstract Most previous approaches to syntactic parsing of Chinese rely on a preprocessing step of word segmentation, thereby assuming there was a clearly defined boundary between morphology and syntax in Chinese. [sent-3, score-0.417]
2 Hence in practice the strict separation of morphology and syntax in the Chinese language proves to be untenable. [sent-5, score-0.22]
3 We present a unified dependency parsing approach for Chinese which takes unsegmented sentences as input and outputs both morphological and syntactic structures with a single model and algorithm. [sent-6, score-1.219]
4 By removing the intermediate word segmentation, the unified parser no longer needs separate notions for words and phrases. [sent-7, score-0.459]
5 Evaluation proves the effectiveness of the unified model and algorithm in parsing structures of words, phrases and sen- tences simultaneously. [sent-8, score-0.837]
6 1445 The first argument is that many sub-word linguistic units (such as suffixes and prefixes) are so productive that they can lead to a huge number of outof-vocabulary words for natural language processing systems. [sent-13, score-0.243]
7 Of course, the way out of this dilemma is to parse the internal structures of these words. [sent-22, score-0.465]
8 Meanwhile, pseudo OOVs can be largely eliminated through analyzing their structures, thus greatly facilitating syntactic and semantic analysis of sentences. [sent-24, score-0.223]
9 In fact, previous studies have revealed other good reasons for parsing internal structures of words (Zhao, 2009; Li, 2011). [sent-25, score-0.652]
10 Lc a2n0g1u2ag Aes Psorcoicaetsiosin fgo arn Cdo Cmopmutpauti oantiaoln Lailn Ngautiustriacls MODMOD 环境 NN 保护 NN 法 NN MOD 刑 NN 法 NN Figure 1: Unified parsing of words and phrases. [sent-28, score-0.228]
11 causes lots of incompatible annotations in currently available corpora. [sent-29, score-0.323]
12 These annotations are incompatible since in both cases the character 法 ‘law’ bears exactly the same meaning and usage (e. [sent-32, score-0.352]
13 Incompatibility can be avoided through parsing structures of both words and phrases. [sent-36, score-0.514]
14 A further benefit of unified parsing is to reduce data sparseness. [sent-38, score-0.509]
15 Word structures in Chinese can be excellent guide for parsing phrase structures, and vice versa, due to their similarity. [sent-42, score-0.48]
16 Firstly, in order to get rid of pseudo OOVs and incompatible annotations, we have annotated structures of words in CTB6, after which statistical models can learn structures of words as well as phrases from the augmented treebank (Section 4). [sent-44, score-1.144]
17 Although previous authors have noticed the importance of word-structure parsing (Li, 2011; Zhao, 2009), no detailed description about annotation ofword structures has been provided in the literature. [sent-45, score-0.502]
18 Secondly, we designed a unified dependency parser whose input is unsegmented sentences and its output incorporates both morphological and syntactic structures with a single model and algorithm (Section 5). [sent-46, score-1.171]
19 By removing the intermediate step of word segmentation, our unified parser no longer depends on the unsound notion that there is a clear boundary between words and phrases. [sent-47, score-0.503]
20 Evaluation (Section 6) shows that our unified parser achieves satisfactory accuracies in parsing both morphologi- cal and syntactic structures. [sent-48, score-0.665]
21 2 Pseudo OOVs and Incompatible Annotations In this section we show the surprisingly pervasive nature of pseudo OOVs and incompatible annotations through analysis of five segmented corpora, which are CTB6 and corpus by MSR, PKU, AS and CITYU provided in SIGHAN word segmentation Bakeoffs 2. [sent-52, score-0.629]
22 It’s amazing that for every corpus, over 60% of OOVs are pseudo, meaning they can be avoided if their internal structures were parsed. [sent-55, score-0.462]
23 We then sample 200 word types containing a productive affix from each corpus, and check whether the affix also occurs somewhere else in a phrase, i. [sent-57, score-0.298]
24 3 Incompatible annotations can be removed only through unified parsing of word and phrase structures, as mentioned earlier and illustrated in Figure 1. [sent-63, score-0.622]
25 OBJMOD MODSUBJ 甘肃 NR 省 NN 重视 VV 保险 NN 业 NN Figure 2: Example output of unified dependency parsing of Chinese morphological and syntactic structures. [sent-70, score-0.826]
26 3 Unified Parsing Defined Given an unsegmented sentence 甘 肃 省 重 视保 险 业 ‘Gansu province attaches great importance to insurance industry’, the output of unified dependency parser is shown in Figure 2. [sent-71, score-0.84]
27 As can be seen, this output contains information about word (such as 重 视_VV) as well as phrase structures (such as 重 视_VV 保 险_NN 业_NN), which is what we mean by ‘unified’ parsing. [sent-72, score-0.33]
28 From the perspective of the unified parser, linguistic units are given the same labels as long as they function similarly (e. [sent-75, score-0.354]
29 As a bonus, output of unified parsing incorporates Chinese word segmentation, part-of-speech tagging and dependency parsing. [sent-77, score-0.657]
30 1447 4 Annotation of Word Structures Unified parsing requires a corpus annotated with both morphological and syntactic structures. [sent-81, score-0.417]
31 It only remains for us to annotate internal structures of words in this treebank. [sent-83, score-0.462]
32 1 Scope of Annotation In order to get rid of pseudo OOVs and incompatible annotations, internal structures are annotated for two kinds of words. [sent-85, score-0.872]
33 The first kind contains words with a productive component such as suffix or prefix. [sent-86, score-0.288]
34 Other compound words such as 研 究 ‘research’ have no productive components and are not a cause of pseudo OOVs. [sent-92, score-0.343]
35 Hence their structures are not annotated in the present research. [sent-94, score-0.306]
36 Meanwhile, for single-morpheme words with no structures whatsoever, like 伊拉 克 ‘Iraq’ and 蝙蝠 ‘bat’, annotation of internal structures is of course unnecessary either. [sent-95, score-0.814]
37 Around 80% of sentences contain words whose structures need annotation. [sent-97, score-0.286]
38 2 From Part-of-speeches to Constituents Of all 33 part-of-speech tags in CTB, annotation of word structures is needed for nine tags: NN, VV, JJ, CD, NT, NR, AD, VA and OD. [sent-100, score-0.369]
39 Since part-of-speech tags are preterminals and can only have one terminal word as its child, POS tags of words become constituent labels after annotation of word structures. [sent-101, score-0.28]
40 A fullfledged parser is needed to analyze structures of these words, which incidentally provides us with another motivation for unified morphological and syntactic parsing of Chinese. [sent-113, score-1.028]
41 理 5 Unified Dependency Parsing All previous dependency parsers for Chinese take it for granted that the input sentence is already segmented into words (Li et al. [sent-114, score-0.228]
42 Hence current off-the-shelf algorithms are inadequate for parsing 1448 unsegmented sentences. [sent-117, score-0.375]
43 Instead, a new unified parsing algorithm is given in this section. [sent-118, score-0.509]
44 1 Transitions To map a raw sentence directly to output shown in Figure 2, we define four transitions for the unified dependency parser. [sent-120, score-0.607]
45 They act on a stack containing the incremental parsing results, and a queue holding the incoming Chinese characters of the sentence: SHIFT: the first character in the queue is shifted into the stack as the start of a new word. [sent-121, score-0.76]
46 We see that these transitions mimic the general arcstandard dependency parsing models. [sent-132, score-0.384]
47 Its sole purpose is to assemble characters into words with no internal structures, such as 西 雅 图 ‘Seattle’ . [sent-135, score-0.257]
48 Thus this transition is the key for removing the need of Chinese word segmentation and parsing unsegmented sentences directly. [sent-136, score-0.551]
49 To also output part-of-speech tags and dependency labels, the transitions above can be augmented accordingly. [sent-137, score-0.331]
50 Also, LEFT ·and RIGHT should be augmented with appropriate dependency relations, such as LEFT·SUBJ for a dependency between verb and subject. [sent-139, score-0.256]
51 Readers interested in implementing their own unified dependency parsers are invited to study this example carefully. [sent-142, score-0.472]
52 2 Model Due to structural ambiguity, there might be quite a lot of possibilities for parsing a given raw sentence. [sent-144, score-0.244]
53 Hence at each step in the parsing process, all four transitions defined above may be applicable. [sent-145, score-0.276]
54 If S denotes the stack holding the partial results, and Q the queue storing the incoming Chinese characters of a raw sentence, then transition-based parsing features are extracted from S and Q according to those feature templates in Table 5. [sent-155, score-0.591]
55 The first beam contains unfinished parsing results, while the second holds completed parses. [sent-166, score-0.353]
56 Note that in line 20 of the algorithm, the beam for completed parsers are pruned at each iteration of the parsing process. [sent-174, score-0.398]
57 Algorithm 1 Beam Search Decoding Algorithm 1Beam Search Decoding 1:candidates ← {STARTITEM()} 2: agenda ← φ 3: completed ← φ 4: loop 5: for all candidate in candidates do 6: for all legal action of candidate do 7: newc ← EXPAND(candidate, action) 8: nife COMPLETED(newc) then 9: completed. [sent-176, score-0.304]
58 Though we annotated structures of words in CTB6, most previously results were on CTB5, a subset of the former treebank. [sent-191, score-0.344]
59 1 Dependency Parsing of Morphological and Syntactic Structures If we look back at the Figure 2, it’s clear that a dependency relation is correctly parsed if and only if three conditions are met: Firstly, words at both ends of the dependency are correctly segmented. [sent-194, score-0.254]
60 1 Results on the Original CTB5 We first train our unified dependency parser with the original treebank CTB5. [sent-202, score-0.567]
61 e, the original CTB5, unified parser performs much better than the result of a pipelined approach reported by Zhang and Clark (201 1). [sent-206, score-0.602]
62 There are about 30% of relative error reduction for the unlabeled dependency parsing results. [sent-207, score-0.298]
63 2 90 Table 8: Evaluation results on CTB5 with word structures annotated. [sent-212, score-0.248]
64 2 Results on CTB with Structures of Words Annotated Then we train the parser with CTB5 augmented with our annotations of internal structures of words. [sent-216, score-0.637]
65 For purpose of better comparability, we report results on both the original dependencies of CTB5 and on the dependencies of CTB5 plus those of the internal structures of words. [sent-217, score-0.522]
66 First, note that compared to another result by Zhang and Nivre (201 1), whose input were sentences with gold standard word segmentation and POS tags, our F-score is only slightly lower even with input of unsegmented sentences. [sent-219, score-0.361]
67 This is understandable since gold-standard segmentation and POS tags greatly reduced the uncertainty of parsing results. [sent-220, score-0.423]
68 For the unified parser, the improvement ofF-score from 79. [sent-221, score-0.319]
69 Internal structures of words are be mostly local in comparison with phrase and sentence structures. [sent-227, score-0.328]
70 Therefore, with the addition of word structures, the overall dependency parsing accuracy naturally can be improved. [sent-228, score-0.298]
71 2 Chinese Word Segmentation From the example in Figure 2, it is clear that output of unified parser contains Chinese word segmentation information. [sent-230, score-0.637]
72 5807 Table 9: Word segmentation results of our parser and the best performance reported in literature on the same dataset. [sent-234, score-0.278]
73 For maximal comparability, we train the unified parser on the original CTB5 data used by previous studies. [sent-245, score-0.421]
74 Despite the fact that the performance of our unified parser does not exceed the best reported result so far, which probably might be caused by some minute implementation specific details, it’s fair to say that our parser performs at the level of state-of-the-art in Chinese word segmentation. [sent-247, score-0.523]
75 3 Joint Word Segmentation and POS Tagging From Figure 2 we see that besides word segmentation, output of the unified parser also includes partof-speech tags. [sent-249, score-0.461]
76 The results are in Table 10, in which for ease of comparison, again we train the unified parser with the vanilla version of CTB5. [sent-252, score-0.421]
77 We can see that unified parser performs at virtually the same level of accuracy compared with previous best systems. [sent-253, score-0.421]
78 7 Related Work Researchers have noticed the necessity of parsing the internal structures of words in Chinese. [sent-254, score-0.652]
79 Li (201 1) gave an method that could take raw sentences as input and output phrase structures and internal structures of words. [sent-255, score-0.808]
80 Secondly, we specify the details of annotating structures of words, with the annotations being made publicly available. [sent-261, score-0.319]
81 Zhao (2009) presented a dependency parser which regards each Chinese character as a word and then analyzes the dependency relations between characters, using ordinary dependency parsing algorithms. [sent-262, score-0.685]
82 e, unified parsing of Chinese word and phrase structures, bears a deeper meaning. [sent-274, score-0.587]
83 As demonstrated in Section 1 and by Li (201 1), structures of words and phrases usually have significant similarity, and the distinction between them is very difficult to define, even for expert linguists. [sent-275, score-0.332]
84 But for real world applications, such subtle matters can safely be ignored if we could analyzed morphological and syntactic structures in a unified framework. [sent-276, score-0.736]
85 What applications really cares is structures instead of whether a linguistic unit is a word or phrase. [sent-277, score-0.248]
86 Another notable line of research closely related to the present work is to annotate and parse the flat structures of noun phrases (NP) (Vadas and Curran, 2007; Vadas and Curran, 2011). [sent-278, score-0.381]
87 g, nouns, verbs, adverbs, adjectives etc) whose structures are not annotated by CTB, and whose presence could cause lots of pseudo OOVs and incompatible annotations. [sent-281, score-0.727]
88 Second, the problem we are trying to solve is a crucial observation specific to Chinese language, that is, in lots of cases forcing a separation of words and phrases leads to awkward situations for NLP systems. [sent-282, score-0.262]
89 Remember that in Section 2 we demonstrated that all corpora we examined had the problem of pseudo OOVs and incompatible annotations. [sent-283, score-0.345]
90 In comparison, the problem Vadas and Curran (2007) tried to solve is a lack of annotation for structures of NPs in currently available treebanks, or to put it in another way, a problem more closely related to treebanks rather than certain languages. [sent-284, score-0.312]
91 8 Discussion and Conclusion Chinese word segmentation is an indispensable step for traditional approaches to syntactic parsing of Chinese. [sent-285, score-0.42]
92 g, parsing) left to higher level structures of phrases and sentences. [sent-287, score-0.294]
93 This difficulty cannot be left unheeded, as we have shown quantitatively that in practice it causes lots of real troubles such as too many OOVs and incompatible annotations. [sent-289, score-0.252]
94 We showed how these undesirable consequences can be resolved by annotation of the internal structures of words, and by unified parsing of morphological and syntactic structures in Chinese. [sent-290, score-1.414]
95 Unified parsing of morphological and syntactic structures of Chinese can also be implemented with a pipelined approach, in which we first segment input sentences into words or affixes (i. [sent-291, score-0.894]
96 However, a unified parsing approach using a single model as presented in this 1453 paper offers several advantages over pipelined approaches. [sent-293, score-0.69]
97 Secondly, both the unified model and the algorithm are conceptually much more simpler than pipelined approaches. [sent-295, score-0.5]
98 A single generative model for joint morphological segmentation and syntactic parsing. [sent-320, score-0.383]
99 A discriminative model for joint morphological disambiguation and dependency parsing. [sent-342, score-0.261]
100 A stacked sub-word model for joint chinese word segmentation and part-of-speech tagging. [sent-358, score-0.394]
wordName wordTfidf (topN-words)
[('unified', 0.319), ('oovs', 0.318), ('structures', 0.248), ('parsing', 0.19), ('unsegmented', 0.185), ('pipelined', 0.181), ('chinese', 0.18), ('internal', 0.176), ('segmentation', 0.176), ('incompatible', 0.176), ('pseudo', 0.169), ('productive', 0.136), ('nn', 0.118), ('morphological', 0.115), ('queue', 0.114), ('vadas', 0.113), ('dependency', 0.108), ('ctb', 0.107), ('newc', 0.106), ('kruengkrai', 0.103), ('parser', 0.102), ('morphology', 0.096), ('stack', 0.096), ('completed', 0.092), ('transitions', 0.086), ('suffix', 0.081), ('lots', 0.076), ('agenda', 0.071), ('zhang', 0.071), ('annotations', 0.071), ('beam', 0.071), ('character', 0.069), ('affixes', 0.068), ('clark', 0.067), ('law', 0.067), ('constituent', 0.064), ('annotation', 0.064), ('affix', 0.062), ('zhao', 0.061), ('annotated', 0.058), ('separation', 0.057), ('tags', 0.057), ('templates', 0.056), ('pos', 0.056), ('syntactic', 0.054), ('raw', 0.054), ('topb', 0.053), ('append', 0.05), ('dependencies', 0.049), ('li', 0.048), ('association', 0.046), ('flat', 0.046), ('nr', 0.046), ('phrases', 0.046), ('insurance', 0.045), ('rid', 0.045), ('awkward', 0.045), ('parsers', 0.045), ('boundary', 0.044), ('characters', 0.043), ('curran', 0.042), ('shift', 0.042), ('phrase', 0.042), ('oregon', 0.041), ('parse', 0.041), ('province', 0.041), ('beams', 0.041), ('yue', 0.041), ('augmented', 0.04), ('secondly', 0.04), ('course', 0.04), ('output', 0.04), ('perceptron', 0.039), ('words', 0.038), ('industry', 0.038), ('comparability', 0.038), ('zhongguo', 0.038), ('somewhere', 0.038), ('holding', 0.038), ('avoided', 0.038), ('treebank', 0.038), ('joint', 0.038), ('decoding', 0.037), ('np', 0.037), ('segmented', 0.037), ('portland', 0.036), ('bears', 0.036), ('thirdly', 0.036), ('succeeds', 0.036), ('hence', 0.035), ('units', 0.035), ('technologies', 0.035), ('action', 0.035), ('suffixes', 0.034), ('vv', 0.034), ('proves', 0.034), ('syntax', 0.033), ('kind', 0.033), ('seattle', 0.032), ('gen', 0.032)]
simIndex simValue paperId paperTitle
same-paper 1 1.0000001 131 emnlp-2012-Unified Dependency Parsing of Chinese Morphological and Syntactic Structures
Author: Zhongguo Li ; Guodong Zhou
Abstract: Most previous approaches to syntactic parsing of Chinese rely on a preprocessing step of word segmentation, thereby assuming there was a clearly defined boundary between morphology and syntax in Chinese. We show how this assumption can fail badly, leading to many out-of-vocabulary words and incompatible annotations. Hence in practice the strict separation of morphology and syntax in the Chinese language proves to be untenable. We present a unified dependency parsing approach for Chinese which takes unsegmented sentences as input and outputs both morphological and syntactic structures with a single model and algorithm. By removing the intermediate word segmentation, the unified parser no longer needs separate notions for words and phrases. Evaluation proves the effectiveness of the unified model and algorithm in parsing structures of words, phrases and sen- tences simultaneously. 1
2 0.28873643 70 emnlp-2012-Joint Chinese Word Segmentation, POS Tagging and Parsing
Author: Xian Qian ; Yang Liu
Abstract: In this paper, we propose a novel decoding algorithm for discriminative joint Chinese word segmentation, part-of-speech (POS) tagging, and parsing. Previous work often used a pipeline method Chinese word segmentation followed by POS tagging and parsing, which suffers from error propagation and is unable to leverage information in later modules for earlier components. In our approach, we train the three individual models separately during training, and incorporate them together in a unified framework during decoding. We extend the CYK parsing algorithm so that it can deal with word segmentation and POS tagging features. As far as we know, this is the first work on joint Chinese word segmentation, POS tagging and parsing. Our experimental results on Chinese Tree Bank 5 corpus show that our approach outperforms the state-of-the-art pipeline system. –
Author: Bernd Bohnet ; Joakim Nivre
Abstract: Most current dependency parsers presuppose that input words have been morphologically disambiguated using a part-of-speech tagger before parsing begins. We present a transitionbased system for joint part-of-speech tagging and labeled dependency parsing with nonprojective trees. Experimental evaluation on Chinese, Czech, English and German shows consistent improvements in both tagging and parsing accuracy when compared to a pipeline system, which lead to improved state-of-theart results for all languages.
4 0.18154867 106 emnlp-2012-Part-of-Speech Tagging for Chinese-English Mixed Texts with Dynamic Features
Author: Jiayi Zhao ; Xipeng Qiu ; Shu Zhang ; Feng Ji ; Xuanjing Huang
Abstract: In modern Chinese articles or conversations, it is very popular to involve a few English words, especially in emails and Internet literature. Therefore, it becomes an important and challenging topic to analyze Chinese-English mixed texts. The underlying problem is how to tag part-of-speech (POS) for the English words involved. Due to the lack of specially annotated corpus, most of the English words are tagged as the oversimplified type, “foreign words”. In this paper, we present a method using dynamic features to tag POS of mixed texts. Experiments show that our method achieves higher performance than traditional sequence labeling methods. Meanwhile, our method also boosts the performance of POS tagging for pure Chinese texts.
5 0.16263509 68 emnlp-2012-Iterative Annotation Transformation with Predict-Self Reestimation for Chinese Word Segmentation
Author: Wenbin Jiang ; Fandong Meng ; Qun Liu ; Yajuan Lu
Abstract: In this paper we first describe the technology of automatic annotation transformation, which is based on the annotation adaptation algorithm (Jiang et al., 2009). It can automatically transform a human-annotated corpus from one annotation guideline to another. We then propose two optimization strategies, iterative training and predict-selfreestimation, to further improve the accuracy of annotation guideline transformation. Experiments on Chinese word segmentation show that, the iterative training strategy together with predictself reestimation brings significant improvement over the simple annotation transformation baseline, and leads to classifiers with significantly higher accuracy and several times faster processing than annotation adaptation does. On the Penn Chinese Treebank 5.0, , it achieves an F-measure of 98.43%, significantly outperforms previous works although using a single classifier with only local features.
6 0.13909617 57 emnlp-2012-Generalized Higher-Order Dependency Parsing with Cube Pruning
7 0.11029223 105 emnlp-2012-Parser Showdown at the Wall Street Corral: An Empirical Investigation of Error Types in Parser Output
8 0.10289442 66 emnlp-2012-Improving Transition-Based Dependency Parsing with Buffer Transitions
9 0.10245813 126 emnlp-2012-Training Factored PCFGs with Expectation Propagation
10 0.093640387 28 emnlp-2012-Collocation Polarity Disambiguation Using Web-based Pseudo Contexts
11 0.093275324 136 emnlp-2012-Weakly Supervised Training of Semantic Parsers
12 0.091666222 64 emnlp-2012-Improved Parsing and POS Tagging Using Inter-Sentence Consistency Constraints
13 0.089504138 81 emnlp-2012-Learning to Map into a Universal POS Tagset
14 0.089014575 7 emnlp-2012-A Novel Discriminative Framework for Sentence-Level Discourse Analysis
15 0.088936746 45 emnlp-2012-Exploiting Chunk-level Features to Improve Phrase Chunking
16 0.077666894 99 emnlp-2012-On Amortizing Inference Cost for Structured Prediction
17 0.074584275 38 emnlp-2012-Employing Compositional Semantics and Discourse Consistency in Chinese Event Extraction
18 0.074277498 65 emnlp-2012-Improving NLP through Marginalization of Hidden Syntactic Structure
19 0.071636558 127 emnlp-2012-Transforming Trees to Improve Syntactic Convergence
20 0.071633279 124 emnlp-2012-Three Dependency-and-Boundary Models for Grammar Induction
topicId topicWeight
[(0, 0.286), (1, -0.212), (2, 0.239), (3, -0.088), (4, 0.171), (5, -0.0), (6, -0.003), (7, -0.091), (8, -0.158), (9, -0.112), (10, -0.119), (11, -0.108), (12, 0.038), (13, -0.081), (14, 0.136), (15, 0.043), (16, 0.03), (17, 0.118), (18, -0.036), (19, -0.056), (20, -0.116), (21, 0.072), (22, -0.091), (23, 0.042), (24, -0.086), (25, -0.037), (26, 0.031), (27, -0.013), (28, 0.027), (29, 0.1), (30, 0.076), (31, 0.027), (32, 0.007), (33, -0.085), (34, 0.035), (35, -0.019), (36, 0.008), (37, 0.003), (38, -0.025), (39, 0.032), (40, 0.017), (41, -0.034), (42, -0.117), (43, -0.032), (44, -0.021), (45, 0.006), (46, 0.001), (47, 0.01), (48, 0.059), (49, -0.017)]
simIndex simValue paperId paperTitle
same-paper 1 0.96645856 131 emnlp-2012-Unified Dependency Parsing of Chinese Morphological and Syntactic Structures
Author: Zhongguo Li ; Guodong Zhou
Abstract: Most previous approaches to syntactic parsing of Chinese rely on a preprocessing step of word segmentation, thereby assuming there was a clearly defined boundary between morphology and syntax in Chinese. We show how this assumption can fail badly, leading to many out-of-vocabulary words and incompatible annotations. Hence in practice the strict separation of morphology and syntax in the Chinese language proves to be untenable. We present a unified dependency parsing approach for Chinese which takes unsegmented sentences as input and outputs both morphological and syntactic structures with a single model and algorithm. By removing the intermediate word segmentation, the unified parser no longer needs separate notions for words and phrases. Evaluation proves the effectiveness of the unified model and algorithm in parsing structures of words, phrases and sen- tences simultaneously. 1
2 0.82833451 70 emnlp-2012-Joint Chinese Word Segmentation, POS Tagging and Parsing
Author: Xian Qian ; Yang Liu
Abstract: In this paper, we propose a novel decoding algorithm for discriminative joint Chinese word segmentation, part-of-speech (POS) tagging, and parsing. Previous work often used a pipeline method Chinese word segmentation followed by POS tagging and parsing, which suffers from error propagation and is unable to leverage information in later modules for earlier components. In our approach, we train the three individual models separately during training, and incorporate them together in a unified framework during decoding. We extend the CYK parsing algorithm so that it can deal with word segmentation and POS tagging features. As far as we know, this is the first work on joint Chinese word segmentation, POS tagging and parsing. Our experimental results on Chinese Tree Bank 5 corpus show that our approach outperforms the state-of-the-art pipeline system. –
Author: Bernd Bohnet ; Joakim Nivre
Abstract: Most current dependency parsers presuppose that input words have been morphologically disambiguated using a part-of-speech tagger before parsing begins. We present a transitionbased system for joint part-of-speech tagging and labeled dependency parsing with nonprojective trees. Experimental evaluation on Chinese, Czech, English and German shows consistent improvements in both tagging and parsing accuracy when compared to a pipeline system, which lead to improved state-of-theart results for all languages.
4 0.6974439 106 emnlp-2012-Part-of-Speech Tagging for Chinese-English Mixed Texts with Dynamic Features
Author: Jiayi Zhao ; Xipeng Qiu ; Shu Zhang ; Feng Ji ; Xuanjing Huang
Abstract: In modern Chinese articles or conversations, it is very popular to involve a few English words, especially in emails and Internet literature. Therefore, it becomes an important and challenging topic to analyze Chinese-English mixed texts. The underlying problem is how to tag part-of-speech (POS) for the English words involved. Due to the lack of specially annotated corpus, most of the English words are tagged as the oversimplified type, “foreign words”. In this paper, we present a method using dynamic features to tag POS of mixed texts. Experiments show that our method achieves higher performance than traditional sequence labeling methods. Meanwhile, our method also boosts the performance of POS tagging for pure Chinese texts.
5 0.58660436 45 emnlp-2012-Exploiting Chunk-level Features to Improve Phrase Chunking
Author: Junsheng Zhou ; Weiguang Qu ; Fen Zhang
Abstract: Most existing systems solved the phrase chunking task with the sequence labeling approaches, in which the chunk candidates cannot be treated as a whole during parsing process so that the chunk-level features cannot be exploited in a natural way. In this paper, we formulate phrase chunking as a joint segmentation and labeling task. We propose an efficient dynamic programming algorithm with pruning for decoding, which allows the direct use of the features describing the internal characteristics of chunk and the features capturing the correlations between adjacent chunks. A relaxed, online maximum margin training algorithm is used for learning. Within this framework, we explored a variety of effective feature representations for Chinese phrase chunking. The experimental results show that the use of chunk-level features can lead to significant performance improvement, and that our approach achieves state-of-the-art performance. In particular, our approach is much better at recognizing long and complicated phrases. 1
6 0.56819683 68 emnlp-2012-Iterative Annotation Transformation with Predict-Self Reestimation for Chinese Word Segmentation
7 0.5615437 57 emnlp-2012-Generalized Higher-Order Dependency Parsing with Cube Pruning
8 0.50311452 66 emnlp-2012-Improving Transition-Based Dependency Parsing with Buffer Transitions
9 0.50154161 38 emnlp-2012-Employing Compositional Semantics and Discourse Consistency in Chinese Event Extraction
10 0.49234053 64 emnlp-2012-Improved Parsing and POS Tagging Using Inter-Sentence Consistency Constraints
11 0.45365065 7 emnlp-2012-A Novel Discriminative Framework for Sentence-Level Discourse Analysis
12 0.39644814 46 emnlp-2012-Exploiting Reducibility in Unsupervised Dependency Parsing
13 0.38165328 126 emnlp-2012-Training Factored PCFGs with Expectation Propagation
14 0.37928185 88 emnlp-2012-Minimal Dependency Length in Realization Ranking
15 0.36698031 48 emnlp-2012-Exploring Adaptor Grammars for Native Language Identification
16 0.34427804 119 emnlp-2012-Spectral Dependency Parsing with Latent Variables
17 0.34394735 127 emnlp-2012-Transforming Trees to Improve Syntactic Convergence
18 0.34174922 105 emnlp-2012-Parser Showdown at the Wall Street Corral: An Empirical Investigation of Error Types in Parser Output
19 0.33615589 124 emnlp-2012-Three Dependency-and-Boundary Models for Grammar Induction
20 0.33101502 81 emnlp-2012-Learning to Map into a Universal POS Tagset
topicId topicWeight
[(2, 0.017), (16, 0.058), (25, 0.013), (34, 0.068), (41, 0.313), (45, 0.012), (47, 0.02), (60, 0.127), (63, 0.049), (64, 0.024), (65, 0.027), (70, 0.033), (73, 0.016), (74, 0.051), (76, 0.061), (80, 0.011), (86, 0.018), (95, 0.022)]
simIndex simValue paperId paperTitle
same-paper 1 0.79947209 131 emnlp-2012-Unified Dependency Parsing of Chinese Morphological and Syntactic Structures
Author: Zhongguo Li ; Guodong Zhou
Abstract: Most previous approaches to syntactic parsing of Chinese rely on a preprocessing step of word segmentation, thereby assuming there was a clearly defined boundary between morphology and syntax in Chinese. We show how this assumption can fail badly, leading to many out-of-vocabulary words and incompatible annotations. Hence in practice the strict separation of morphology and syntax in the Chinese language proves to be untenable. We present a unified dependency parsing approach for Chinese which takes unsegmented sentences as input and outputs both morphological and syntactic structures with a single model and algorithm. By removing the intermediate word segmentation, the unified parser no longer needs separate notions for words and phrases. Evaluation proves the effectiveness of the unified model and algorithm in parsing structures of words, phrases and sen- tences simultaneously. 1
2 0.70929557 11 emnlp-2012-A Systematic Comparison of Phrase Table Pruning Techniques
Author: Richard Zens ; Daisy Stanton ; Peng Xu
Abstract: When trained on very large parallel corpora, the phrase table component of a machine translation system grows to consume vast computational resources. In this paper, we introduce a novel pruning criterion that places phrase table pruning on a sound theoretical foundation. Systematic experiments on four language pairs under various data conditions show that our principled approach is superior to existing ad hoc pruning methods.
3 0.56073552 42 emnlp-2012-Entropy-based Pruning for Phrase-based Machine Translation
Author: Wang Ling ; Joao Graca ; Isabel Trancoso ; Alan Black
Abstract: Phrase-based machine translation models have shown to yield better translations than Word-based models, since phrase pairs encode the contextual information that is needed for a more accurate translation. However, many phrase pairs do not encode any relevant context, which means that the translation event encoded in that phrase pair is led by smaller translation events that are independent from each other, and can be found on smaller phrase pairs, with little or no loss in translation accuracy. In this work, we propose a relative entropy model for translation models, that measures how likely a phrase pair encodes a translation event that is derivable using smaller translation events with similar probabilities. This model is then applied to phrase table pruning. Tests show that considerable amounts of phrase pairs can be excluded, without much impact on the transla- . tion quality. In fact, we show that better translations can be obtained using our pruned models, due to the compression of the search space during decoding.
4 0.50847244 114 emnlp-2012-Revisiting the Predictability of Language: Response Completion in Social Media
Author: Bo Pang ; Sujith Ravi
Abstract: The question “how predictable is English?” has long fascinated researchers. While prior work has focused on formal English typically used in news articles, we turn to texts generated by users in online settings that are more informal in nature. We are motivated by a novel application scenario: given the difficulty of typing on mobile devices, can we help reduce typing effort with message completion, especially in conversational settings? We propose a method for automatic response completion. Our approach models both the language used in responses and the specific context provided by the original message. Our experimental results on a large-scale dataset show that both components help reduce typing effort. We also perform an information-theoretic study in this setting and examine the entropy of user-generated content, especially in con- versational scenarios, to better understand predictability of user generated English.
5 0.50069529 22 emnlp-2012-Automatically Constructing a Normalisation Dictionary for Microblogs
Author: Bo Han ; Paul Cook ; Timothy Baldwin
Abstract: Microblog normalisation methods often utilise complex models and struggle to differentiate between correctly-spelled unknown words and lexical variants of known words. In this paper, we propose a method for constructing a dictionary of lexical variants of known words that facilitates lexical normalisation via simple string substitution (e.g. tomorrow for tmrw). We use context information to generate possible variant and normalisation pairs and then rank these by string similarity. Highlyranked pairs are selected to populate the dictionary. We show that a dictionary-based approach achieves state-of-the-art performance for both F-score and word error rate on a standard dataset. Compared with other methods, this approach offers a fast, lightweight and easy-to-use solution, and is thus suitable for high-volume microblog pre-processing. 1 Lexical Normalisation A staggering number of short text “microblog” messages are produced every day through social media such as Twitter (Twitter, 2011). The immense volume of real-time, user-generated microblogs that flows through sites has been shown to have utility in applications such as disaster detection (Sakaki et al., 2010), sentiment analysis (Jiang et al., 2011; Gonz a´lez-Ib ´a n˜ez et al., 2011), and event discovery (Weng and Lee, 2011; Benson et al., 2011). However, due to the spontaneous nature of the posts, microblogs are notoriously noisy, containing many non-standard forms e.g., tmrw “tomorrow” and 2day “today” which degrade the performance of — — 421 natural language processing (NLP) tools (Ritter et al., 2010; Han and Baldwin, 2011). To reduce this effect, attempts have been made to adapt NLP tools to microblog data (Gimpel et al., 2011; Foster et al., 2011; Liu et al., 2011b; Ritter et al., 2011). An alternative approach is to pre-normalise non-standard lexical variants to their standard orthography (Liu et al., 2011a; Han and Baldwin, 2011; Xue et al., 2011; Gouws et al., 2011). For example, se u 2morw!!! would be normalised to see you tomorrow! The normalisation approach is especially attractive as a preprocessing step for applications which rely on keyword match or word frequency statistics. For example, earthqu, eathquake, and earthquakeee all attested in a Twitter corpus have the standard form earthquake; by normalising these types to their standard form, better coverage can be achieved for keyword-based methods, and better word frequency estimates can be obtained. In this paper, we focus on the task of lexical normalisation of English Twitter messages, in which out-of-vocabulary (OOV) tokens are normalised to their in-vocabulary (IV) standard form, i.e., a standard form that is in a dictionary. Following other recent work on lexical normalisation (Liu et al., 2011a; Han and Baldwin, 2011; Gouws et al., 2011; Liu et al., 2012), we specifically focus on one-to-one normalisation in which one OOV token is normalised to one IV word. Naturally, not all OOV words in microblogs are lexical variants of IV words: named entities, e.g., — — are prevalent in microblogs, but not all named entities are included in our dictionary. One challenge for lexical normalisation is therefore to disPLraoncge uadgineg Lse oafr tnhineg 2,0 p1a2g Jeosin 42t C1–o4n3f2e,re Jnecjue Iosnla Enmd,p Kiroicraela, M 1e2t–h1o4ds Ju ilny N 20a1tu2r.a ?lc L2a0n1g2ua Agseso Pcrioactieosnsi fnogr a Cnodm Cpoumtaptiuotna tilo Lnianlg Nuaist uircasl tinguish those OOV tokens that require normalisation from those that are well-formed. Recent unsupervised approaches have not attempted to distinguish such tokens from other types of OOV tokens (Cook and Stevenson, 2009; Liu et al., 2011a), limiting their applicability to real-world normalisation tasks. Other approaches (Han and Baldwin, 2011; Gouws et al., 2011) have followed a cascaded approach in which lexical variants are first identified, and then normalised. However, such two-step approaches suffer from poor lexical variant identification performance, which is propagated to the normalisation step. Motivated by the observation that most lexical variants have an unambiguous standard form (especially for longer tokens), and that a lexical variant and its standard form typically occur in similar contexts, in this paper we propose methods for automatically constructing a lexical normalisation dictionary a dictionary whose entries consist — of (lexical variant, standard form) pairs that enables type-based normalisation. Despite the simplicity of this dictionary-based normalisation method, we show it to outperform previously-proposed approaches. This very fast, lightweight solution is suitable for real-time processing of the large volume of streaming microblog data available from Twitter, and offers a simple solution to the lexical variant detection problem that hinders other normalisation methods. Furthermore, this dictionary-based method can be easily integrated with other more-complex normalisation approaches (Liu et al., 2011a; Han and Baldwin, 2011; Gouws et al., 2011) to produce hybrid systems. After discussing related work in Section 2, we present an overview of our dictionary-based approach to normalisation in Section 3. In Sections 4 and 5 we experimentally select the optimised context similarity parameters and string similarity reranking method. We present experimental results on the unseen test data in Section 6, and offer some concluding remarks in Section 7. — 2 Related Work Given a token t, lexical normalisation is the task of finding arg max P(s|t) ∝ arg max P(t| s)P(s), wofh efinred s igs tahreg smtaanxdaPrd(s form, i.e., an aIVx Pw(otr|sd). PSt(asn)-, dardly in lexical normalisation, t is assumed to be an 422 OOV token, relative to a fixed dictionary. In practice, not all OOV tokens should be normalised; i.e., only lexical variants (e.g., tmrw “tomorrow”) should be normalised and tokens that are OOV but otherwise not lexical variants (e.g., iPad “iPad”) should be unchanged. Most work in this area focuses only on the normalisation task itself, oftentimes assuming that the task of lexical variant detection has already been completed. Various approaches have been proposed to estimate the error model, P(t|s). For example, in work on spell-checking, eBl,ril Pl (atn|ds) M. Fooorre e (2000) improve on a standard edit-distance approach by considering multi-character edit operations; Toutanova and Moore (2002) build on this by incorporating phonological information. Li et al. (2006) utilise distributional similarity (Lin, 1998) to correct misspelled search queries. In text message normalisation, Choudhury et al. (2007) model the letter transformations and emissions using a hidden Markov model (Rabiner, 1989). Cook and Stevenson (2009) and Xue et al. (201 1) propose multiple simple error models, each of which captures a particular way in which lexical variants are formed, such as phonetic spelling (e.g., epik “epic”) or clipping (e.g., walkin “walking”). Nevertheless, optimally weighting the various error models in these approaches is challenging. Without pre-categorising lexical variants into different types, Liu et al. (201 1a) collect Google search snippets from carefully-designed queries from which they then extract noisy lexical variant– standard form pairs. These pairs are used to train a conditional random field (Lafferty et al., 2001) to estimate P(t|s) at the character level. One shortcoming eo fP querying a ese cahracrha engine teol. .o Obtanein strhaoirnt-ing pairs is it tends to be costly in terms of time and bandwidth. Here we exploit microblog data directly to derive (lexical variant, standard form) pairs, instead of relying on external resources. In morerecent work, Liu et al. (2012) endeavour to improve the accuracy of top-n normalisation candidates by integrating human cognitive inference, characterlevel transformations and spell checking in their normalisation model. The encouraging results shift the focus to reranking and promoting the correct normalisation to the top-1 position. However, like much previous work on lexical normalisation, this work assumes perfect lexical variant detection. Aw et al. (2006) and Kaufmann and Kalita (2010) consider normalisation as a machine translation task from lexical variants to standard forms using off-theshelf tools. These methods do not assume that lexical variants have been pre-identified; however, these methods do rely on large quantities of labelled training data, which is not available for microblogs. Recently, Han and Baldwin (201 1) and Gouws et al. (201 1) propose two-step unsupervised approaches to normalisation, in which lexical variants are first identified, and then normalised. They approach lexical variant detection by using a context fitness classifier (Han and Baldwin, 2011) or through dictionary lookup (Gouws et al., 2011). However, the lexical variant detection of both meth- ods is rather unreliable, indicating the challenge of this aspect of normalisation. Both of these approaches incorporate a relatively small normalisation dictionary to capture frequent lexical variants with high precision. In particular, Gouws et al. (201 1) produce a small normalisation lexicon based on distributional similarity and string similarity (Lodhi et al., 2002). Our method adopts a similar strategy using distributional/string similarity, but instead of constructing a small lexicon for preprocessing, we build a much wider-coverage normalisation dictionary and opt for a fully lexiconbased end-to-end normalisation approach. In contrast to the normalisation dictionaries of Han and Baldwin (201 1) and Gouws et al. (201 1) which focus on very frequent lexical variants, we focus on moderate frequency lexical variants of a minimum character length, which tend to have unambiguous standard forms; our intention is to produce normalisation lexicons that are complementary to those currently available. Furthermore, we investigate the impact of a variety of contextual and string similarity measures on the quality of the resulting lexicons. In summary, our dictionary-based normalisation ap- proach is a lightweight end-to-end method which performs both lexical variant detection and normalisation, and thus is suitable for practical online preprocessing, despite its simplicity. 423 3 A Lexical Normalisation Dictionary Before discussing our method for creating a normalisation dictionary, we first discuss the feasibility of such an approach. 3.1 Feasibility Dictionary lookup approaches to normalisation have been shown to have high precision but low recall (Han and Baldwin, 2011; Gouws et al., 2011). Frequent (lexical variant, standard form) pairs such as (u, you) are typically included in the dictionaries used by such methods, while less-frequent items such as (g0tta, gotta) are generally omitted. Because of the degree of lexical creativity and large number of non-standard forms observed on Twitter, a wide-coverage normalisation dictionary would be expensive to construct manually. Based on the assumption that lexical variants occur in similar con- texts to their standard forms, however, it should be possible to automatically construct a normalisation dictionary with wider coverage than is currently available. Dictionary lookup is a type-based approach to normalisation, i.e., every token instance of a given type will always be normalised in the same way. However, lexical variants can be ambiguous, e.g., y corresponds to “you” in yeah, y r right! LOL but “why” in AM CONFUSED!!! y you did that? Nevertheless, the relative occurrence of ambiguous lexical variants is small (Liu et al., 2011a), and it has been observed that while shorter variants such as y are often ambiguous, longer variants tend to be unambiguous. For example bthday and 4eva are unlikely to have standard forms other than “birthday” and “forever”, respectively. Therefore, the normalisation lexicons we produce will only contain entries for OOVs with character length greater than a specified threshold, which are likely to have an unambiguous standard form. 3.2 Overview of approach Our method for constructing a normalisation dictio- nary is as follows: Input: Tokenised English tweets 1. Extract (OOV, IV) pairs based on distributional similarity. 2. Re-rank the extracted pairs by string similarity. Output: A list of (OOV, IV) pairs ordered by string similarity; select the top-n pairs for inclusion in the normalisation lexicon. In Step 1, we leverage large volumes of Twitter data to identify the most distributionally-similar IV type for each OOV type. The result of this process is a set of (OOV, IV) pairs, ranked by distributional similarity. The extracted pairs will include (lexical variant, standard form) pairs, such as (tmrw, tomorrow), but will also contain false positives such as (Tusday, Sunday) Tusday is a lexical variant, but its standard form is not “Sunday” and (Youtube, web) Youtube is an OOV named entity, not a lexical variant. Nevertheless, lexical variants are typically formed from their standard forms through regular processes (Thurlow, 2003) e.g., the omission of characters and from this perspective Sunday and web are not plausible standard — — — — — forms for Tusday and Youtube, respectively. In Step 2, we therefore capture this intuition to re-rank the extracted pairs by string similarity. The top-n items in this re-ranked list then form the normalisation lexicon, which is based only on development data. Although computationally-expensive to build, this dictionary can be created offline. Once built, it then offers a very fast approach to normalisation. We can only reliably compute distributional similarity for types that are moderately frequent in a corpus. Nevertheless, many lexical variants are sufficiently frequent to be able to compute distributional similarity, and can potentially make their way into our normalisation lexicon. This approach is not suitable for normalising low-frequency lexical variants, nor is it suitable for shorter lexical variant types which as discussed in Section 3.1 are more likely to have an ambiguous standard form. Nevertheless, previously-proposed normalisation methods that can handle such phenomena also rely in part on a normalisation lexicon. The normalisation lexicons we create can therefore be easily integrated with previous approaches to form hybrid normalisation systems. — — 4 Contextually-similar Pair Generation Our objective is to extract contextually-similar (OOV, IV) pairs from a large-scale collection of mi424 croblog data. Fundamentally, the surrounding words define the primary context, but there are different ways of representing context and different similarity measures we can use, which may influence the quality of generated normalisation pairs. In representing the context, we experimentally explore the following factors: (1) context window size (from 1 to 3 tokens on both sides); (2) n-gram order ofthe context tokens (unigram, bigram, trigram); (3) whether context words are indexed for relative position or not; and (4) whether we use all context tokens, or only IV words. Because high-accuracy linguistic processing tools for Twitter are still under exploration (Liu et al., 2011b; Gimpel et al., 2011; Ritter et al., 2011; Foster et al., 2011), we do not consider richer representations of context, for example, incorporating information about part-of-speech tags or syntax. We also experiment with a number of simple but widely-used geometric and information theoretic distance/similarity measures. In particular, we use Kullback–Leibler (KL) divergence (Kullback and Leibler, 195 1), Jensen–Shannon (JS) divergence (Lin, 1991), Euclidean distance and Cosine distance. We use a corpus of 10 million English tweets to do parameter tuning over, and a larger corpus of tweets in the final candidate ranking. All tweets were collected from September 2010 to January 2011 via the Twitter API.1 From the raw data we extract English tweets using a language identification tool (Lui and Baldwin, 2011), and then apply a simplified Twitter tokeniser (adapted from O’Connor et al. (2010)). We use the Aspell dictionary (v6.06)2 to determine whether a word is IV, and only include in our normalisation dictionary OOV tokens with at least 64 occurrences in the corpus and character length ≥ 4, both of which were determined through empirical 4o,b bsoetrhva otifo wnh. Fcohr w weearceh d OetOeVrm winoedrd t type ginh the corpus, we select the most similar IV type to form (OOV, IV) pairs. To further narrow the search space, we only consider IV words which are morphophonemically similar to the OOV type, follow- ing settings in Han and Baldwin (201 1).3 1http s : / / dev .twitter . com/ docs / st reaming-api /methods 2http : / / aspe l .net / l 3We only consider IV words within an edit distance of 2 or a phonemic edit distance of 1from the OOV type, and we further In order to evaluate the generated pairs, we randomly selected 1000 OOV words from the 10 million tweet corpus. We set up an annotation task on Amazon Mechanical Turk,4 presenting five independent annotators with each word type (with no context) and asking for corrections where appropriate. For instance, given tmrw, the annotators would likely identify it as a non-standard variant of “tomorrow”. For correct OOV words like iPad, on the other hand, we would expect them to leave the word unchanged. If 3 or more of the 5 annotators make the same suggestion (in the form of either a canonical spelling or leaving the word unchanged), we include this in our gold standard for evaluation. In total, this resulted in 351 lexical variants and 282 correct OOV words, accounting for 63.3% of the 1000 OOV words. These 633 OOV words were used as (OOV, IV) pairs for parameter tuning. The remainder of the 1000 OOV words were ignored on the grounds that there was not sufficient consensus amongst the annotators.5 Contextually-similar pair generation aims to include as many correct normalisation pairs as possible. We evaluate the quality of the normalisation pairs using “Cumulative Gain” (CG): XN0 CG = Xreli0 Xi=1 Suppose there are N0 correct generated pairs (oovi, ivi), each of which is weighted by reli0, the frequency of oovi to indicate its relative importance; for example, (thinkin, thinking) has a higher weight than (g0tta, gotta) because thinkin is more frequent than g0tta in our corpus. In this evaluation we don’t consider the position of normalisation pairs, and nor do we penalise incorrect pairs. Instead, we push distinguishing between correct and incorrect pairs into the downstream re-ranking step in which we incorporate string similarity information. Given the development data and CG, we run an exhaustive search of parameter combinations over only consider the top 30% most-frequent of these IV words. 4https : / /www .mturk .com/mturk/welcome 5Note that the objective of this annotation task is to identify lexical variants that have agreed-upon standard forms irrespective of context, as a special case of the more general task of lexical normalisation (where context may or may not play a significant role in the determination of the normalisation). 425 our development corpus. The five best parameter combinations are shown in Table 1. We notice the CG is almost identical for the top combinations. As a context window size of 3 incurs a heavy processing and memory overhead over a size of 2, we use the 3rd-best parameter combination for subsequent experiments, namely: context window of ±2 tokens, teoxkpeenr bigrams, positional index, nadnodw wK oLf divergence as our distance measure. To better understand the sensitivity of the method to each parameter, we perform a post-hoc parameter analysis relative to a default setting (as underlined in Table 2), altering one parameter at a time. The results in Table 2 show that bigrams outperform other n-gram orders by a large margin (note that the evaluation is based on a log scale), and information-theoretic measures are superior to the geometric measures. Furthermore, it also indicates using the positional indexing better captures context. However, there is little to distinguish context modelling with just IV words or all tokens. Similarly, the context window size has relatively little impact on the overall performance, supporting our earlier observation from Table 1. 5 Pair Re-ranking by String Similarity Once the contextually-similar (OOV, IV) pairs are generated using the selected parameters in Section 4, we further re-rank this set of pairs in an attempt to boost morphophonemically-similar pairs like (bananaz, bananas), and penalise noisy pairs like (paninis, beans). Instead of using the small 10 million tweet corpus, from this step onwards, we use a larger corpus of 80 million English tweets (collected over the same period as the development corpus) to develop a larger-scale normalisation dictionary. This is because once pairs are generated, re-ranking based on string comparison is much faster. We only include in the dictionary OOV words with a token frequency > 15 to include more OOV types than in Section 4, and again apply a minimum length cutoff of 4 char- acters. To measure how well our re-ranking method promotes correct pairs and demotes incorrect pairs (including both OOV words that should not be normalised, e.g. (Youtube, web), and incorrect normalRankWindow sizen-gramPositional index?Lex. choiceSim/distance measurelog(CG) 1±32YesAllKL divergence19.571 2 ±±33 2 No All KL divergence 19.562 3 ±±23 2 Yes All KL divergence 19.562 4 ±±32 2 Yes IVs KL divergence 19.561 5 ±±23 2 Yes IVs JS divergence 19.554 ±2 Table 1: The five best parameter combinations in the exhaustive search of parameter combinations Window sizen-gramPositional index?Lexical choiceSimilarity/distance measure ±1 19.3251 19.328Yes 19.328IVs 19.335KL divergence 19.328 ±±21 1199..332275 2 19.571 No 19.263 All 19.328 Euclidean 19.227 ±±32 1199..332287 3 19.324 JS divergence 19.31 1 Cosine 19.170 Table 2: Parameter sensitivity analysis measured as log(CG) for correctly-generated pairs. We tune one parameter at a time, using the default (underlined) setting for other parameters; the non-exhaustive best-performing setting in each case is indicated in bold. isations for lexical variants, e.g. (bcuz, cause)), we modify our evaluation metric from Section 4 to evaluate the ranking at different points, using Discounted Cumulative Gain (DCG@N: Jarvelin and Kekalainen (2002)): DCG@N = rel1+XiN=2logr2el(i ) where reli again represents the frequency of the OOV, but it can be gain (a positive number) or loss (a negative number), depending on whether the ith pair is correct or incorrect. Because we also expect correct pairs to be ranked higher than incorrect pairs, DCG@N takes both factors into account. Given the generated pairs and the evaluation metric, we first consider three baselines: no re-ranking (i.e., the final ranking is that of the contextual similarity scores), and re-rankings of the pairs based on the frequencies of the OOVs in the Twitter corpus, and the IV unigram frequencies in the Google Web 1T corpus (Brants and Franz, 2006) to get less-noisy frequency estimates. We also compared a variety of re-rankings based on a number of string similarity measures that have been previously considered in normalisation work (reviewed in Section 2). We experiment with standard edit distance (Levenshtein, 1966), edit distance over double metaphone codes (phonetic edit distance: (Philips, 2000)), longest common subsequence ratio over the consonant edit distance of the paired words (hereafter, denoted as 426 consonant edit distance: (Contractor et al., 2010)), and a string subsequence kernel (Lodhi et al., 2002). In Figure 1, we present the DCG@N results for each of our ranking methods at different rank cutoffs. Ranking by OOV frequency is motivated by the assumption that lexical variants are frequently used by social media users. This is confirmed by our findings that lexical pairs like (goin, going) and (nite, night) are at the top of the ranking. However, many proper nouns and named entities are also used frequently and ranked at the top, mixed with lexical variants like (Facebook, speech) and (Youtube, web). In ranking by IV word frequency, we assume the lexical variants are usually derived from frequently-used IV equivalents, e.g. (abou, about). However, many less-frequent lexical variant types have high-frequency (IV) normalisations. For instance, the highest-frequency IV word the has more than 40 OOV lexical variants, such as tthe and thhe. These less-frequent types occupy the top positions, reducing the cumulative gain. Compared with these two baselines, ranking by default contextual similarity scores delivers promising results. It successfully ranks many more intuitive normalisation pairs at the top, such as (2day, today) and (wknd, weekend), but also ranks some incorrect pairs highly, such as (needa, gotta). The string similarity-based methods perform better than our baselines in general. Through manual analysis, we found that standard edit distance ranking is fairly accurate for lexical variants with low edit distance to their standard forms, but fails to identify heavily-altered variants like (tmrw, tomorrow). Consonant edit distance is similar to standard edit distance, but places many longer words at the top of the ranking. Edit distance over double metaphone codes (phonetic edit distance) performs particularly well for lexical variants that include character repetitions commonly used for emphasis on Twitter because such repetitions do not typically alter the phonetic codes. Compared with the other methods, the string subsequence kernel delivers encouraging results. It measures common character subsequences of length n between (OOV, IV) pairs. Because it is computationally expensive to calculate similarity for larger n, we choose n=2, following Gouws et al. (201 1). As N (the lexicon size cut-off) increases, the performance drops more slowly than the other meth— — ods. Although this method fails to rank heavilyaltered variants such as (4get,forget) highly, it typically works well for longer words. Given that we focus on longer OOVs (specifically those longer than 4 characters), this ultimately isn’t a great handicap. 6 Evaluation Given the re-ranked pairs from Section 5, here we apply them to a token-level normalisation task using the normalisation dataset of Han and Baldwin (201 1). 6.1 Metrics We evaluate using the standard evaluation metrics of precision (P), recall (R) and F-score (F) as detailed below. We also consider the false alarm rate (FA) and word error rate (WER), also as shown below. FA measures the negative effects of applying normalisation; a good approach to normalisation should not (incorrectly) normalise tokens that are already in their standard form and do not require normalisation.6 WER, like F-score, shows the overall benefits of normalisation, but unlike F-score, measures how many token-level edits are required for the output to be the same as the ground truth data. In general, dictionaries with a high F-score/low WER and low FA 6FA + P ≤ 1because some lexical variants might be incorrectly Ano +rm Pa ≤lise 1d b. 427 are preferable. P = R= F = FA = WER = # cor#re nctolrym naolrismedal tioskeden toskens # to ckoernresc rtelyqu niori nmga nloisremda tloiskaetniosn P2P +R R # inco#rr encotrlmya nliosremda tloikseedns tokens # token edits n#ee adlletd o akfetnesr normalisation 6.2 Results We select the three best re-ranking methods, and best cut-off N for each method, based on the highest DCG@N value for a given method over the development data, as presented in Figure 1. Namely, they are string subsequence kernel (S-dict, N=40,000), double metaphone edit distance (DMdict, N=10,000) and default contextual similarity without re-ranking (C-dict, N=10,000).7 We evaluate each of the learned dictionaries in Table 3. We also compare each dictionary with the performance of the manually-constructed Internet slang dictionary (HB-dict) used by Han and Baldwin (201 1), the small automatically-derived dictionary of Gouws et al. (201 1) (GHM-dict), and combinations of the different dictionaries. In addition, the contribution of these dictionaries in hybrid normalisation approaches is also presented, in which we first normalise OOVs using a given dictionary (combined or otherwise), and then apply the normalisation method of Gouws et al. (201 1) based on consonant edit distance (GHM-norm), or the approach of Han and Baldwin (201 1) based on the summation of many unsupervised approaches (HB-norm), to the remaining OOVs. Results are shown in Table 3, and discussed below. 6.2.1 Individual Dictionaries Overall, the individual dictionaries derived by the re-ranking methods (DM-dict, S-dict) perform bet- 7We also experimented with combining ranks using Mean Reciprocal Rank. However, the combined rank didn’t improve performance on the development data. We plan to explore other ranking aggregation methods in future work. 1 3 5 7 9 11 31 51 71 91 N cut−offs Figure 1: Re-ranking based on different string similarity methods. ter than that based on contextual similarity (C-dict) in terms of precision and false alarm rate, indicating the importance of re-ranking. Even though C-dict delivers higher recall indicating that many lexical variants are correctly normalised this is offset by its high false alarm rate, which is particularly undesirable in normalisation. Because S-dict has better performance than DM-dict in terms of both F-score and WER, and a much lower false alarm rate than C-dict, subsequent results are presented using S-dict only. — — Both HB-dict and GHM-dict achieve better than 90% precision with moderate recall. Compared to these methods, S-dict is not competitive in terms of either precision or recall. This result seems rather discouraging. However, considering that S-dict is an automatically-constructed dictionary targeting lexical variants of varying frequency, it is not surprising that the precision is worse than that of HB-dict which is manually-constructed and GHM-dict which includes entries only for more-frequent OOVs for which distributional similarity is more accurate. Additionally, the recall of S-dict is hampered by the — — — 428 restriction on lexical variant token length of 4 characters. 6.2.2 Combined Dictionaries Next we look to combining HB-dict, GHM-dict and S-dict. In combining the dictionaries, a given OOV word can be listed with different standard forms in different dictionaries. In such cases we use the following preferences for dictionaries motivated by our confidence in the normalisation pairs — of the dictionaries to resolve conflicts: HB-dict > GHM-dict > S-dict. When we combine dictionaries in the second section of Table 3, we find that they contain complementary information: in each case the recall and F-score are higher for the combined dictionary than any of the individual dictionaries. The combination of HB-dict+GHM-dict produces only a small improvement in terms of F-score over HBdict (the better-performing dictionary) suggesting that, as claimed, HB-dict and GHM-dict share many frequent normalisation pairs. HB-dict+S-dict and GHM-dict+S-dict, on the other hand, improve sub— MethodPrecisionRecallF-ScoreFalse AlarmWord Error Rate C-dict0.4740.2180.2990.2980.103 DM-dict S-dict HB-dict GHM-dict 0.727 0.700 0.915 0.982 0.106 0.179 0.435 0.319 0.185 0.285 0.590 0.482 0.145 0.162 0.048 0.000 0.102 0.097 0.066 0.076 HB-dict+S-dict0.8400.6010.7010.0900.052 GHM-dict+S-dict HB-dict+GHM-dict HB-dict+GHM-dict+S-dict 0.863 0.920 0.847 0.498 0.465 0.630 0.632 0.618 0.723 0.072 0.045 0.086 0.061 0.063 0.049 GHM-dict+GHM-norm0.3380.5780.4270.4580.135 HB-dict+GHM-dict+S-dict+GHM-norm HB-dict+HB-norm HB-dict+GHM-dict+S-dict+HB-norm 0.406 0.515 0.527 0.715 0.771 0.789 0.518 0.618 0.632 0.468 0.332 0.332 0.124 0.081 0.079 Table 3: Normalisation results using our derived dictionaries (contextual similarity (C-dict); double metaphone rendering (DM-dict); string subsequence kernel scores (S-dict)), the dictionary of Gouws et al. (201 1) (GHM-dict), the Internet slang dictionary (HB-dict) from Han and Baldwin (201 1), and combinations of these dictionaries. In addition, we combine the dictionaries with the normalisation method of Gouws et al. (201 1) (GHM-norm) and the combined unsupervised approach of Han and Baldwin (201 1) (HB-norm). stantially over HB-dict and GHM-dict, respectively, indicating that S-dict contains markedly different entries to both HB-dict and GHM-dict. The best Fscore and WER are obtained using the combination of all three dictionaries, HB-dict+GHM-dict+S-dict. Furthermore, the difference between the results using HB-dict+GHM-dict+S-dict and HB-dict+GHMdict is statistically significant (p < 0.01), based on the computationally-intensive Monte Carlo method of Yeh (2000), demonstrating the contribution of Sdict. 6.2.3 Hybrid Approaches The methods of Gouws et al. (201 1) (i.e. GHM-dict+GHM-norm) and Han and Baldwin (201 1) (i.e. HB-dict+HB-norm) have lower precision and higher false alarm rates than the dictionarybased approaches; this is largely caused by lexical variant detection errors.8 Using all dictionaries in combination with these methods HB-dict+GHM-dict+S-dict+GHM-norm and HBdict+GHM-dict+S-dict+HB-norm gives some improvements, but the false alarm rates remain high. Despite the limitations of a pure dictionary-based approach to normalisation discussed in Section 3.1 the current best practical approach to normal— — — — 8Here we report results that do not assume perfect detection of lexical variants, unlike the original published results in each case. 429 Error typeOOVDSitcat.ndard fGoromld (a) pluralsplayeplayersplayer (b) negation unlike like dislike (c) possessives anyones anyone anyone ’s (d) correct OOVs iphone phone iphone (e) test data errors durin during durin (f) ambiguity siging signing singing Table 4: Error types in the combined dictionary (HBdict+GHM-dict+S-dict) isation is to use a lexicon, combining hand-built and automatically-learned normalisation dictionaries. 6.3 Discussion and Error Analysis We first manually analyse the errors in the combined dictionary (HB-dict+GHM-dict+S-dict) and give examples of each error type in Table 4. The most frequent word errors are caused by slight morphologi- cal variations, including plural forms (a), negations (b), possessive cases (c), and OOVs that are correct and do not require normalisation (d). In addition, we also notice some missing annotations where lexical variants are skipped by human annotations but captured by our method (e). Ambiguity (f) definitely exists in longer OOVs, however, these cases do not appear to have a strong negative impact on the normalisation performance. An example of a remainLength cut-off (N)#VariantsPrecisionRecall (≥ N)Recall (all)False Alarm ≥45560.700Rec0al.l3 8(≥1 N)0.1790.162 ≥≥54 382 0.814 0.471 0.152 0.122 ≥≥65 254 0.804 0.484 0.104 0.131 ≥≥76 138 0.793 0.471 0.055 0.122 ≥71380.7930.4710.0550.122 Table 5: S-dict normalisation results broken down according to OOV token length. Recall is presented both over the subset of instances of length ≥ N in the data (“Recall (≥ N)”), and over the entirety of the dataset (“Recall (all)”); “su#bVsaertia onftis n” sitsa tnhcee snu omfb leenrg othf t≥ok Nen iinns tthaenc deast ao f( “tRhee cinadllic (a≥ted N length idn o othveer rt tehset d eanttaisreetty. ing miscellaneous error is bday “birthday”, which is mis-normalised as day. To further study the influence of OOV word length relative to the normalisation performance, we conduct a fine-grained analysis of the performance of the derived dictionary (S-dict) in Table 5, broken down across different OOV word lengths. The results generally support our hypothesis that our method works better for longer OOV words. The derived dictionary is much more reliable for longer tokens (length 5, 6, and 7 characters) in terms of precision and false alarm. Although the recall is relatively modest, in the future we intend to improve recall by mining more normalisation pairs from larger collections of microblog data. 7 Conclusions and Future Work In this paper, we describe a method for automatically constructing a normalisation dictionary that supports normalisation of microblog text through direct substitution of lexical variants with their standard forms. After investigating the impact of different distributional and string similarity methods on the quality of the dictionary, we present experimental results on a standard dataset showing that our proposed methods acquire high quality (lexical variant, standard form) pairs, with reasonable coverage, and achieve state-of-the-art end-toend lexical normalisation performance on a realworld token-level task. Furthermore, this dictionarylookup method combines the detection and normalisation of lexical variants into a simple, lightweight solution which is suitable for processing of highvolume microblog feeds. In the future, we intend to improve our dictionary by leveraging the constantly-growing volume of microblog data, and considering alternative ways to combine distributional and string similarity. In addi430 tion to direct evaluation, we also want to explore the benefits of applying normalisation for downstream social media text processing applications, e.g. event detection. Acknowledgements We would like to thank the three anonymous reviewers for their insightful comments, and Stephan Gouws for kindly sharing his data and discussing his work. NICTA is funded by the Australian government as represented by Department of Broadband, Communication and Digital Economy, and the Australian Research Council through the ICT centre of Excellence programme. References AiTi Aw, Min Zhang, Juan Xiao, and Jian Su. 2006. A phrase-based statistical model for SMS text normalization. In Proceedings of COLING/ACL 2006, pages 33–40, Sydney, Australia. Edward Benson, Aria Haghighi, and Regina Barzilay. 2011. Event discovery in social media feeds. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 389–398, Portland, Oregon, USA. Thorsten Brants and Alex Franz. 2006. Web 1T 5-gram Version 1. Eric Brill and Robert C. Moore. 2000. An improved error model for noisy channel spelling correction. In Proceedings of the 38th Annual Meeting of the Association for Computational Linguistics, pages 286–293, Hong Kong. Monojit Choudhury, Rahul Saraf, Vijit Jain, Animesh Mukherjee, Sudeshna Sarkar, and Anupam Basu. 2007. Investigation and modeling of the structure of texting language. International Journal on Document Analysis and Recognition, 10: 157–174. Danish Contractor, Tanveer A. Faruquie, and L. Venkata Subramaniam. 2010. Unsupervised cleansing of noisy text. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING 2010), pages 189–196, Beijing, China. Paul Cook and Suzanne Stevenson. 2009. An unsu- pervised model for text message normalization. In CALC ’09: Proceedings of the Workshop on Computational Approaches to Linguistic Creativity, pages 71– 78, Boulder, USA. Jennifer Foster, O¨zlem C ¸etinoglu, Joachim Wagner, Joseph L. Roux, Stephen Hogan, Joakim Nivre, Deirdre Hogan, and Josef van Genabith. 2011. #hardtoparse: POS Tagging and Parsing the Twitterverse. In Analyzing Microtext: Papers from the 2011 AAAI Workshop, volume WS-1 1-05 of AAAI Workshops, pages 20–25, San Francisco, CA, USA. Kevin Gimpel, Nathan Schneider, Brendan O’Connor, Dipanjan Das, Daniel Mills, Jacob Eisenstein, Michael Heilman, Dani Yogatama, Jeffrey Flanigan, and Noah A. Smith. 2011. Part-of-speech tagging for Twitter: Annotation, features, and experiments. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 42–47, Portland, Oregon, USA. Roberto Gonz a´lez-Ib ´a n˜ez, Smaranda Muresan, and Nina Wacholder. 2011. Identifying sarcasm in Twitter: a closer look. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 581–586, Portland, Oregon, USA. Stephan Gouws, Dirk Hovy, and Donald Metzler. 2011. Unsupervised mining of lexical variants from noisy text. In Proceedings of the First workshop on Unsupervised Learning in NLP, pages 82–90, Edinburgh, Scotland, UK. Bo Han and Timothy Baldwin. 2011. Lexical normalisation of short text messages: Makn sens a #twitter. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 368–378, Portland, Oregon, USA. K. Jarvelin and J. Kekalainen. 2002. Cumulated gainbased evaluation of IR techniques. ACM Transactions on Information Systems, 20(4). Long Jiang, Mo Yu, Ming Zhou, Xiaohua Liu, and Tiejun Zhao. 2011. Target-dependent Twitter sentiment classification. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 151–160, Portland, Oregon, USA. Joseph Kaufmann and Jugal Kalita. 2010. Syntactic normalization of Twitter messages. In International Con431 ference on Natural Language Processing, Kharagpur, India. S. Kullback and R. A. Leibler. 1951. On information and sufficiency. Annals of Mathematical Statistics, 22:49– 86. John D. Lafferty, Andrew McCallum, and Fernando C. N. Pereira. 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In Proceedings ofthe Eighteenth International Conference on Machine Learning, pages 282–289, San Francisco, CA, USA. Vladimir I. Levenshtein. 1966. Binary codes capable of correcting deletions, insertions, and reversals. Soviet Physics Doklady, 10:707–710. Mu Li, Yang Zhang, Muhua Zhu, and Ming Zhou. 2006. Exploring distributional similarity based models for query spelling correction. In Proceedings of COLING/ACL 2006, pages 1025–1032, Sydney, Australia. Jianhua Lin. 1991. Divergence measures based on the shannon entropy. IEEE Transactions on Information Theory, 37(1): 145–151. Dekang Lin. 1998. Automatic retrieval and clustering of similar words. In Proceedings of the 36th Annual Meeting of the ACL and 1 International Con7th ference on Computational Linguistics (COLING/ACL98), pages 768–774, Montreal, Quebec, Canada. Fei Liu, Fuliang Weng, Bingqing Wang, and Yang Liu. 2011a. Insertion, deletion, or substitution? normalizing text messages without pre-categorization nor supervision. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 71–76, Portland, Oregon, USA. Xiaohua Liu, Shaodian Zhang, Furu Wei, and Ming Zhou. 2011b. Recognizing named entities in tweets. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-HLT 2011), pages 359–367, Portland, Oregon, USA. Fei Liu, Fuliang Weng, and Xiao Jiang. 2012. A broadcoverage normalization system for social media language. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics (ACL 2012), Jeju, Republic of Korea. Huma Lodhi, Craig Saunders, John Shawe-Taylor, Nello Cristianini, and Chris Watkins. 2002. Text classification using string kernels. J. Mach. Learn. Res., 2:419– 444. Marco Lui and Timothy Baldwin. 2011. Cross-domain feature selection for language identification. In Proceedings of the 5th International Joint Conference on Natural Language Processing (IJCNLP 2011), pages 553–561, Chiang Mai, Thailand. Brendan O’Connor, Michel Krieger, and David Ahn. 2010. TweetMotif: Exploratory search and topic summarization for Twitter. In Proceedings of the 4th International Conference on Weblogs and Social Media (ICWSM 2010), pages 384–385, Washington, USA. Lawrence Philips. 2000. The double metaphone search algorithm. C/C++ Users Journal, 18:38–43. Lawrence R. Rabiner. 1989. A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77(2):257–286. Alan Ritter, Colin Cherry, and Bill Dolan. 2010. Unsupervised modeling of Twitter conversations. In Proceedings of Human Language Technologies: The 11th Annual Conference of the North American Chap- ter of the Association for Computational Linguistics (NAACL-HLT 2010), pages 172–180, Los Angeles, USA. Alan Ritter, Sam Clark, Mausam, and Oren Etzioni. 2011. Named entity recognition in tweets: An experimental study. In Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing (EMNLP 2011), pages 1524–1534, Edinburgh, Scotland, UK. Takeshi Sakaki, Makoto Okazaki, and Yutaka Matsuo. 2010. Earthquake shakes Twitter users: real-time event detection by social sensors. In Proceedings of the 19th International Conference on the World Wide Web (WWW 2010), pages 851–860, Raleigh, North Carolina, USA. Crispin Thurlow. 2003. Generation txt? The sociolinguistics of young people’s text-messaging. Discourse Analysis Online, 1(1). Kristina Toutanova and Robert C. Moore. 2002. Pronunciation modeling for improved spelling correction. In Proceedings of the 40th Annual Meeting of the ACL and 3rd Annual Meeting of the NAACL (ACL-02), pages 144–15 1, Philadelphia, USA. Official Blog Twitter. 2011. 200 million tweets per day. Retrived at August 17th, 2011. Jianshu Weng and Bu-Sung Lee. 2011. Event detection in Twitter. In Proceedings of the 5th International Conference on Weblogs and Social Media (ICWSM 2011), Barcelona, Spain. Zhenzhen Xue, Dawei Yin, and Brian D. Davison. 2011. Normalizing microtext. In Proceedings of the AAAI11 Workshop on Analyzing Microtext, pages 74–79, San Francisco, USA. Alexander Yeh. 2000. More accurate tests for the statistical significance of result differences. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING 2010), pages 947–953, Saarbr¨ ucken, Germany. 432
6 0.49616188 136 emnlp-2012-Weakly Supervised Training of Semantic Parsers
7 0.49562633 123 emnlp-2012-Syntactic Transfer Using a Bilingual Lexicon
9 0.49291182 70 emnlp-2012-Joint Chinese Word Segmentation, POS Tagging and Parsing
10 0.48853296 71 emnlp-2012-Joint Entity and Event Coreference Resolution across Documents
11 0.48713657 92 emnlp-2012-Multi-Domain Learning: When Do Domains Matter?
12 0.48668215 24 emnlp-2012-Biased Representation Learning for Domain Adaptation
13 0.48408428 23 emnlp-2012-Besting the Quiz Master: Crowdsourcing Incremental Classification Games
14 0.48407689 45 emnlp-2012-Exploiting Chunk-level Features to Improve Phrase Chunking
15 0.48323467 122 emnlp-2012-Syntactic Surprisal Affects Spoken Word Duration in Conversational Contexts
16 0.48202446 14 emnlp-2012-A Weakly Supervised Model for Sentence-Level Semantic Orientation Analysis with Multiple Experts
17 0.48025218 124 emnlp-2012-Three Dependency-and-Boundary Models for Grammar Induction
18 0.47976199 107 emnlp-2012-Polarity Inducing Latent Semantic Analysis
19 0.47863623 51 emnlp-2012-Extracting Opinion Expressions with semi-Markov Conditional Random Fields
20 0.4778721 39 emnlp-2012-Enlarging Paraphrase Collections through Generalization and Instantiation