acl acl2010 acl2010-9 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Lei Cui ; Dongdong Zhang ; Mu Li ; Ming Zhou ; Tiejun Zhao
Abstract: In hierarchical phrase-based SMT systems, statistical models are integrated to guide the hierarchical rule selection for better translation performance. Previous work mainly focused on the selection of either the source side of a hierarchical rule or the target side of a hierarchical rule rather than considering both of them simultaneously. This paper presents a joint model to predict the selection of hierarchical rules. The proposed model is estimated based on four sub-models where the rich context knowledge from both source and target sides is leveraged. Our method can be easily incorporated into the practical SMT systems with the log-linear model framework. The experimental results show that our method can yield significant improvements in performance.
Reference: text
sentIndex sentText sentNum sentScore
1 com Abstract In hierarchical phrase-based SMT systems, statistical models are integrated to guide the hierarchical rule selection for better translation performance. [sent-8, score-1.12]
2 Previous work mainly focused on the selection of either the source side of a hierarchical rule or the target side of a hierarchical rule rather than considering both of them simultaneously. [sent-9, score-1.511]
3 This paper presents a joint model to predict the selection of hierarchical rules. [sent-10, score-0.443]
4 The proposed model is estimated based on four sub-models where the rich context knowledge from both source and target sides is leveraged. [sent-11, score-0.311]
5 Our method can be easily incorporated into the practical SMT systems with the log-linear model framework. [sent-12, score-0.033]
6 1 Introduction Hierarchical phrase-based model has strong expression capabilities of translation knowledge. [sent-14, score-0.129]
7 It can not only maintain the strength of phrase translation in traditional phrase-based models (Koehn et al. [sent-15, score-0.096]
8 , 2006), but also characterize the complicated long distance reordering similar to syntactic based statistical machine translation (SMT) models (Yamada and Knight, 2001 ; Quirk et al. [sent-17, score-0.145]
9 In hierarchical phrase-based SMT systems, due to the flexibility of rule matching, a huge number of hierarchical rules could be automatically learnt from bilingual training corpus (Chiang, 2005). [sent-24, score-0.948]
10 proper rule selection for hypothesis generation, including both source-side rule selection and targetside rule selection where the source-side rule determines what part of source words to be translated and the target-side rule provides one of the candidate translations of the source-side rule. [sent-26, score-2.587]
11 There is some related work about the hierarchical rule selection. [sent-28, score-0.561]
12 In the original work (Chiang, 2005), the target-side rule selection is analogous to the model in traditional phrase-based SMT system such as Pharaoh (Koehn et al. [sent-29, score-0.595]
13 , 2008) integrate rich context information of non-terminals to predict the target-side rule selection. [sent-33, score-0.483]
14 Different from the above work where the probability distribution of source-side rule selection is uniform, (Setiawan et al. [sent-34, score-0.593]
15 , 2009) proposes to select sourceside rules based on the captured function words which often play an important role in word reordering. [sent-35, score-0.155]
16 There is also some work considering to involve more rich contexts to guide the source-side rule selection. [sent-36, score-0.47]
17 , 2009) explore the source syntactic information to reward exact matching structure rules or punish crossing structure rules. [sent-38, score-0.172]
18 All the previous work mainly focused on either source-side rule selection task or target-side rule selection task rather than both of them together. [sent-39, score-1.124]
19 In this paper, we propose to integrate both source-side and target-side rule selection in a unified model. [sent-41, score-0.562]
20 The intuition is that the joint selection of source-side and target-side rules is more reliable as it conducts the search in a larger space than the single selection task does. [sent-42, score-0.465]
21 It is expected that these two kinds of selection can help and affect each other, which may potentially lead to better hierarchical rule selections with a relative global optimum instead of a local optimum that might be reached in the pre6 Uppsala,P Srwoecde end,i n1g1s-1 o6f Jthuely A 2C0L10 2. [sent-43, score-0.867]
22 Our proposed joint probability model is factored into four sub-models that can be further classified into source-side and targetside rule selection models or context-based and context-free selection models. [sent-46, score-0.923]
23 The context-based models explore rich context features from both source and target sides, including function words, part-of-speech (POS) tags, syntactic structure information and so on. [sent-47, score-0.28]
24 Our model can be easily incorporated as an independent feature into the practical hierarchical phrase-based systems with the log-linear model framework. [sent-48, score-0.255]
25 2 Hierarchical Rule Selection Model Following (Chiang, 2005), hα, γi is used to represFeonltlo a synchronous c 2o0n0t5ex),t hfαre,eγ grammar (SCFG) rule extracted from the training corpus, where α and γ are the source-side and target-side rule respectively. [sent-50, score-0.793]
26 ch Ficoarlrule selection is described as follows: P(α, γ|C) = P(α|C)P(γ|α, C) (1) We decompose the joint probability model into two sub-models based on the Bayes formulation, where the first sub-model is source-side rule selection model and the second one is the target-side rule selection model. [sent-53, score-1.475]
27 CFSM is used to capture how likely the sourceside rule is linguistically motivated or has the corresponding target-side counterpart. [sent-56, score-0.473]
28 For CBSM, it can be naturally viewed as a classification problem where each distinct source-side rule is a single class. [sent-57, score-0.372]
29 However, considering the huge number of classes may cause serious data sparseness problem and thereby degrade the classification accuracy, we approximate CBSM by a binary classification problem which can be solved by the maximum entropy (ME) approach (Berger et al. [sent-58, score-0.111]
30 CBSM estimates the probability of the source-side rule being selected according to the rich context information coming from the surface strings and sub-phrases that will be reduced to non-terminals during decoding. [sent-60, score-0.514]
31 CFTM computes how likely the targetside rule is linguistically motivated, while CBTM predicts how likely the target-side rule is applied according to the clues from the rich context information. [sent-63, score-0.971]
32 1 The acquisition of training instances CBSM and CBTM are trained by ME approach for the binary classification, where a training instance consists of a label and the context related to SCFG rules. [sent-65, score-0.207]
33 The context is divided into source context Figure 1: Example of training instances in CBSM and CBTM. [sent-66, score-0.287]
34 CBSM is trained only based on the source context while CBTM is trained over both the source and the target context. [sent-68, score-0.26]
35 All the training instances are automatically constructed from the bilingual training corpus, which have labels of either positive (i. [sent-69, score-0.293]
36 This section explains how the training instances are constructed for the training of CBSM and CBTM. [sent-74, score-0.194]
37 For the training of CBSM, if the SCFG rule hrs, rti can be extracted based on the rule extracthiron algorithm einx (Chiang, 2005), hυ = 1, C(rs)i itiso cno anlsgtorruicthtemd as a positive instance, o 1t,hCer(wri)sei hυ = 0, C(rs)i is constructed as a negative instance. [sent-76, score-0.917]
38 For the training of CBTM, given rs, suppose there is a SCFG rule set {hrs, rtki |1 ≤ k ≤ n} ethxetrraec itesd a fr SoCmF multiple td {isthrinct sentence pairs nin} the bilingual training corpus, among which we assume hrs , rtii is extracted from the sentence pair hs, ti. [sent-78, score-0.635]
39 For example in Figure 1(c), Rule (1) and Rule (2) are two different SCFG rules extracted from Figure 1(a) and Figure 1(b) respectively, where their source-side rules are the same. [sent-81, score-0.108]
40 CIt( ris noticed)i th araet this instance construction method may lead to a large quantity of negative instances and choke the training procedure. [sent-83, score-0.164]
41 In practice, to limit the size of the training set, the negative instances con- structed based on low-frequency target-side rules are pruned. [sent-84, score-0.218]
42 2 Context-based features for ME training ME approach has the merit of easily combining different features to predict the probability of each class. [sent-86, score-0.156]
43 We incorporate into the ME based model the following informative context-based features to train CBSM and CBTM. [sent-87, score-0.071]
44 These features are carefully designed to reduce the data sparseness problem and some of them are inspired by previous work (He et al. [sent-88, score-0.075]
45 Function word features, which indicate whether the hierarchical source-side/targetside rule strings and sub-phrases covered by non-terminals contain function words that are often important clues of predicting syntactic structures. [sent-94, score-0.637]
46 POS features, which are POS tags of the boundary source words covered by non- terminals. [sent-96, score-0.117]
47 Syntactic features, which are the constituent constraints of hierarchical source-side rules exactly matching or crossing syntactic subtrees. [sent-98, score-0.28]
48 This feature interacts between source and target components since it shows whether the translation ordering is affected. [sent-101, score-0.265]
49 Length features, which are the length of sub-phrases covered by source nonterminals. [sent-103, score-0.117]
50 1 Experiment setting We implement a hierarchical phrase-based system similar to the Hiero (Chiang, 2005) and evaluate our method on the Chinese-to-English translation task. [sent-105, score-0.285]
51 Our bilingual training data comes from FBIS corpus, which consists of around 160K sentence pairs where the source data is parsed by the Berkeley parser (Petrov and Klein, 2007). [sent-106, score-0.194]
52 The ME training toolkit, developed by (Zhang, 2006), is used to train our CBSM and CBTM. [sent-107, score-0.049]
53 The training size of constructed positive instances for both CBSM and CBTM is 4. [sent-108, score-0.18]
54 68M, while the training size of constructed negative instances is 3. [sent-109, score-0.199]
55 The 5-gram language model is trained over the English portion of FBIS corpus plus Xinhua portion of the Gigaword corpus. [sent-117, score-0.033]
56 2 Comparison with related work Our baseline is the implemented Hiero-like SMT system where only the standard features are employed and the performance is state-of-the-art. [sent-123, score-0.038]
57 We compare our method with the baseline and some typical approaches listed in Table 1 where XP+ denotes the approach in (Marton and Resnik, 2008) and TOFW (topological ordering of function words) stands for the method in (Setiawan et al. [sent-124, score-0.038]
58 , 2009)’s work is based on phrasal SMT system with bracketing transduction grammar rules (Wu, 1997) and (Shen et al. [sent-127, score-0.127]
59 01) As shown in Table 1, all the methods outperform the baseline because they have extra models to guide the hierarchical rule selection in some ways which might lead to better translation. [sent-140, score-0.786]
60 Apparently, our method also performs better than the other two approaches, indicating that our method is more effective in the hierarchical rule selection as both source-side and target-side rules are selected together. [sent-141, score-0.805]
61 01) As shown in Table 2, when sub-models are inte9 grated as independent features, the performance is improved compared to the baseline, which shows that each of the sub-models can improve the hierarchical rule selection. [sent-149, score-0.561]
62 It is noticeable that the performance of the source-side rule selection model is comparable with that of the target-side rule selection model. [sent-150, score-1.157]
63 5 Conclusion Hierarchical rule selection is an important and complicated task for hierarchical phrase-based SMT system. [sent-152, score-0.751]
64 We propose a joint probability model for the hierarchical rule selection and the experimental results prove the effectiveness of our approach. [sent-153, score-0.846]
65 In the future work, we will explore more useful features and test our method over the large scale training corpus. [sent-154, score-0.087]
66 A challenge might exist when running the ME training toolkit over a big size of training instances from the large scale training data. [sent-155, score-0.208]
67 the Third Workshop on Statistical Machine Translation, pages 9-17. [sent-185, score-0.045]
wordName wordTfidf (topN-words)
[('cbsm', 0.438), ('rule', 0.372), ('cbtm', 0.27), ('selection', 0.19), ('hierarchical', 0.189), ('setiawan', 0.177), ('cfsm', 0.135), ('cftm', 0.135), ('youhao', 0.135), ('xiong', 0.132), ('smt', 0.128), ('ps', 0.121), ('liu', 0.111), ('scfg', 0.11), ('chiang', 0.109), ('rs', 0.103), ('marton', 0.102), ('qun', 0.102), ('hrs', 0.101), ('sourceside', 0.101), ('interpolation', 0.096), ('shouxun', 0.096), ('translation', 0.096), ('source', 0.081), ('targetside', 0.076), ('hezuo', 0.067), ('rsa', 0.067), ('bilingual', 0.064), ('rich', 0.063), ('shen', 0.062), ('instances', 0.061), ('hendra', 0.059), ('ihi', 0.059), ('zhongjun', 0.059), ('kevin', 0.056), ('pt', 0.055), ('rules', 0.054), ('rta', 0.054), ('negative', 0.054), ('mle', 0.051), ('cui', 0.051), ('gimpel', 0.051), ('harbin', 0.051), ('target', 0.05), ('training', 0.049), ('statistical', 0.049), ('context', 0.048), ('jinxi', 0.048), ('libin', 0.048), ('resnik', 0.048), ('zhang', 0.047), ('deyi', 0.046), ('selections', 0.046), ('topological', 0.046), ('wei', 0.045), ('koehn', 0.045), ('pages', 0.045), ('entropy', 0.043), ('tan', 0.042), ('yuval', 0.042), ('clues', 0.04), ('mi', 0.039), ('acl', 0.039), ('marcu', 0.039), ('features', 0.038), ('ordering', 0.038), ('sparseness', 0.037), ('yamada', 0.037), ('crossing', 0.037), ('fbis', 0.037), ('haizhou', 0.037), ('hs', 0.037), ('transduction', 0.037), ('knight', 0.037), ('berger', 0.036), ('bracketing', 0.036), ('covered', 0.036), ('nist', 0.036), ('sides', 0.036), ('quirk', 0.036), ('constructed', 0.035), ('positive', 0.035), ('guide', 0.035), ('optimum', 0.035), ('side', 0.034), ('decompose', 0.033), ('petrov', 0.033), ('model', 0.033), ('galley', 0.032), ('joint', 0.031), ('yang', 0.031), ('huge', 0.031), ('probability', 0.031), ('ralph', 0.031), ('xu', 0.031), ('zhao', 0.03), ('pi', 0.03), ('della', 0.03), ('aclcoling', 0.03), ('cit', 0.03)]
simIndex simValue paperId paperTitle
same-paper 1 0.99999976 9 acl-2010-A Joint Rule Selection Model for Hierarchical Phrase-Based Translation
Author: Lei Cui ; Dongdong Zhang ; Mu Li ; Ming Zhou ; Tiejun Zhao
Abstract: In hierarchical phrase-based SMT systems, statistical models are integrated to guide the hierarchical rule selection for better translation performance. Previous work mainly focused on the selection of either the source side of a hierarchical rule or the target side of a hierarchical rule rather than considering both of them simultaneously. This paper presents a joint model to predict the selection of hierarchical rules. The proposed model is estimated based on four sub-models where the rich context knowledge from both source and target sides is leveraged. Our method can be easily incorporated into the practical SMT systems with the log-linear model framework. The experimental results show that our method can yield significant improvements in performance.
2 0.26175568 48 acl-2010-Better Filtration and Augmentation for Hierarchical Phrase-Based Translation Rules
Author: Zhiyang Wang ; Yajuan Lv ; Qun Liu ; Young-Sook Hwang
Abstract: This paper presents a novel filtration criterion to restrict the rule extraction for the hierarchical phrase-based translation model, where a bilingual but relaxed wellformed dependency restriction is used to filter out bad rules. Furthermore, a new feature which describes the regularity that the source/target dependency edge triggers the target/source word is also proposed. Experimental results show that, the new criteria weeds out about 40% rules while with translation performance improvement, and the new feature brings an- other improvement to the baseline system, especially on larger corpus.
3 0.20239675 51 acl-2010-Bilingual Sense Similarity for Statistical Machine Translation
Author: Boxing Chen ; George Foster ; Roland Kuhn
Abstract: This paper proposes new algorithms to compute the sense similarity between two units (words, phrases, rules, etc.) from parallel corpora. The sense similarity scores are computed by using the vector space model. We then apply the algorithms to statistical machine translation by computing the sense similarity between the source and target side of translation rule pairs. Similarity scores are used as additional features of the translation model to improve translation performance. Significant improvements are obtained over a state-of-the-art hierarchical phrase-based machine translation system. 1
4 0.19152574 69 acl-2010-Constituency to Dependency Translation with Forests
Author: Haitao Mi ; Qun Liu
Abstract: Tree-to-string systems (and their forestbased extensions) have gained steady popularity thanks to their simplicity and efficiency, but there is a major limitation: they are unable to guarantee the grammaticality of the output, which is explicitly modeled in string-to-tree systems via targetside syntax. We thus propose to combine the advantages of both, and present a novel constituency-to-dependency translation model, which uses constituency forests on the source side to direct the translation, and dependency trees on the target side (as a language model) to ensure grammaticality. Medium-scale experiments show an absolute and statistically significant improvement of +0.7 BLEU points over a state-of-the-art forest-based tree-to-string system even with fewer rules. This is also the first time that a treeto-tree model can surpass tree-to-string counterparts.
5 0.18530589 169 acl-2010-Learning to Translate with Source and Target Syntax
Author: David Chiang
Abstract: Statistical translation models that try to capture the recursive structure of language have been widely adopted over the last few years. These models make use of varying amounts of information from linguistic theory: some use none at all, some use information about the grammar of the target language, some use information about the grammar of the source language. But progress has been slower on translation models that are able to learn the relationship between the grammars of both the source and target language. We discuss the reasons why this has been a challenge, review existing attempts to meet this challenge, and show how some old and new ideas can be combined into a sim- ple approach that uses both source and target syntax for significant improvements in translation accuracy.
6 0.1832197 54 acl-2010-Boosting-Based System Combination for Machine Translation
7 0.15116253 118 acl-2010-Fine-Grained Tree-to-String Translation Rule Extraction
8 0.13620491 115 acl-2010-Filtering Syntactic Constraints for Statistical Machine Translation
9 0.12192748 243 acl-2010-Tree-Based and Forest-Based Translation
10 0.11794072 240 acl-2010-Training Phrase Translation Models with Leaving-One-Out
11 0.11252876 147 acl-2010-Improving Statistical Machine Translation with Monolingual Collocation
12 0.11072469 84 acl-2010-Detecting Errors in Automatically-Parsed Dependency Relations
13 0.10110323 265 acl-2010-cdec: A Decoder, Alignment, and Learning Framework for Finite-State and Context-Free Translation Models
14 0.098692477 133 acl-2010-Hierarchical Search for Word Alignment
15 0.097907804 102 acl-2010-Error Detection for Statistical Machine Translation Using Linguistic Features
16 0.096090861 163 acl-2010-Learning Lexicalized Reordering Models from Reordering Graphs
17 0.094116487 87 acl-2010-Discriminative Modeling of Extraction Sets for Machine Translation
18 0.089021549 249 acl-2010-Unsupervised Search for the Optimal Segmentation for Statistical Machine Translation
19 0.087072104 145 acl-2010-Improving Arabic-to-English Statistical Machine Translation by Reordering Post-Verbal Subjects for Alignment
20 0.086620741 132 acl-2010-Hierarchical Joint Learning: Improving Joint Parsing and Named Entity Recognition with Non-Jointly Labeled Data
topicId topicWeight
[(0, -0.23), (1, -0.233), (2, -0.017), (3, 0.036), (4, -0.025), (5, -0.006), (6, 0.042), (7, 0.013), (8, -0.121), (9, 0.033), (10, 0.131), (11, 0.064), (12, 0.101), (13, -0.034), (14, 0.121), (15, 0.005), (16, -0.007), (17, -0.001), (18, -0.093), (19, 0.006), (20, 0.009), (21, 0.015), (22, -0.004), (23, 0.023), (24, -0.045), (25, 0.058), (26, 0.025), (27, -0.077), (28, 0.131), (29, -0.105), (30, -0.057), (31, 0.007), (32, 0.013), (33, -0.022), (34, 0.141), (35, 0.045), (36, 0.024), (37, 0.124), (38, -0.065), (39, -0.012), (40, 0.068), (41, 0.016), (42, -0.081), (43, -0.075), (44, 0.087), (45, 0.015), (46, 0.035), (47, -0.156), (48, -0.06), (49, -0.069)]
simIndex simValue paperId paperTitle
same-paper 1 0.9694019 9 acl-2010-A Joint Rule Selection Model for Hierarchical Phrase-Based Translation
Author: Lei Cui ; Dongdong Zhang ; Mu Li ; Ming Zhou ; Tiejun Zhao
Abstract: In hierarchical phrase-based SMT systems, statistical models are integrated to guide the hierarchical rule selection for better translation performance. Previous work mainly focused on the selection of either the source side of a hierarchical rule or the target side of a hierarchical rule rather than considering both of them simultaneously. This paper presents a joint model to predict the selection of hierarchical rules. The proposed model is estimated based on four sub-models where the rich context knowledge from both source and target sides is leveraged. Our method can be easily incorporated into the practical SMT systems with the log-linear model framework. The experimental results show that our method can yield significant improvements in performance.
2 0.82810426 48 acl-2010-Better Filtration and Augmentation for Hierarchical Phrase-Based Translation Rules
Author: Zhiyang Wang ; Yajuan Lv ; Qun Liu ; Young-Sook Hwang
Abstract: This paper presents a novel filtration criterion to restrict the rule extraction for the hierarchical phrase-based translation model, where a bilingual but relaxed wellformed dependency restriction is used to filter out bad rules. Furthermore, a new feature which describes the regularity that the source/target dependency edge triggers the target/source word is also proposed. Experimental results show that, the new criteria weeds out about 40% rules while with translation performance improvement, and the new feature brings an- other improvement to the baseline system, especially on larger corpus.
3 0.73900163 118 acl-2010-Fine-Grained Tree-to-String Translation Rule Extraction
Author: Xianchao Wu ; Takuya Matsuzaki ; Jun'ichi Tsujii
Abstract: Tree-to-string translation rules are widely used in linguistically syntax-based statistical machine translation systems. In this paper, we propose to use deep syntactic information for obtaining fine-grained translation rules. A head-driven phrase structure grammar (HPSG) parser is used to obtain the deep syntactic information, which includes a fine-grained description of the syntactic property and a semantic representation of a sentence. We extract fine-grained rules from aligned HPSG tree/forest-string pairs and use them in our tree-to-string and string-to-tree systems. Extensive experiments on largescale bidirectional Japanese-English trans- lations testified the effectiveness of our approach.
4 0.73187262 69 acl-2010-Constituency to Dependency Translation with Forests
Author: Haitao Mi ; Qun Liu
Abstract: Tree-to-string systems (and their forestbased extensions) have gained steady popularity thanks to their simplicity and efficiency, but there is a major limitation: they are unable to guarantee the grammaticality of the output, which is explicitly modeled in string-to-tree systems via targetside syntax. We thus propose to combine the advantages of both, and present a novel constituency-to-dependency translation model, which uses constituency forests on the source side to direct the translation, and dependency trees on the target side (as a language model) to ensure grammaticality. Medium-scale experiments show an absolute and statistically significant improvement of +0.7 BLEU points over a state-of-the-art forest-based tree-to-string system even with fewer rules. This is also the first time that a treeto-tree model can surpass tree-to-string counterparts.
5 0.69202626 51 acl-2010-Bilingual Sense Similarity for Statistical Machine Translation
Author: Boxing Chen ; George Foster ; Roland Kuhn
Abstract: This paper proposes new algorithms to compute the sense similarity between two units (words, phrases, rules, etc.) from parallel corpora. The sense similarity scores are computed by using the vector space model. We then apply the algorithms to statistical machine translation by computing the sense similarity between the source and target side of translation rule pairs. Similarity scores are used as additional features of the translation model to improve translation performance. Significant improvements are obtained over a state-of-the-art hierarchical phrase-based machine translation system. 1
6 0.67477489 169 acl-2010-Learning to Translate with Source and Target Syntax
7 0.64026308 54 acl-2010-Boosting-Based System Combination for Machine Translation
8 0.62139696 265 acl-2010-cdec: A Decoder, Alignment, and Learning Framework for Finite-State and Context-Free Translation Models
9 0.61824 84 acl-2010-Detecting Errors in Automatically-Parsed Dependency Relations
10 0.58927101 243 acl-2010-Tree-Based and Forest-Based Translation
11 0.54495025 119 acl-2010-Fixed Length Word Suffix for Factored Statistical Machine Translation
12 0.54449874 240 acl-2010-Training Phrase Translation Models with Leaving-One-Out
13 0.51733798 115 acl-2010-Filtering Syntactic Constraints for Statistical Machine Translation
14 0.50331378 249 acl-2010-Unsupervised Search for the Optimal Segmentation for Statistical Machine Translation
15 0.50158972 102 acl-2010-Error Detection for Statistical Machine Translation Using Linguistic Features
16 0.49168438 201 acl-2010-Pseudo-Word for Phrase-Based Machine Translation
17 0.46091887 52 acl-2010-Bitext Dependency Parsing with Bilingual Subtree Constraints
18 0.45253178 45 acl-2010-Balancing User Effort and Translation Error in Interactive Machine Translation via Confidence Measures
19 0.45132911 151 acl-2010-Intelligent Selection of Language Model Training Data
20 0.43845996 135 acl-2010-Hindi-to-Urdu Machine Translation through Transliteration
topicId topicWeight
[(14, 0.017), (18, 0.012), (20, 0.026), (25, 0.124), (33, 0.017), (39, 0.015), (42, 0.018), (59, 0.14), (73, 0.039), (78, 0.018), (83, 0.093), (95, 0.231), (98, 0.141)]
simIndex simValue paperId paperTitle
same-paper 1 0.82664335 9 acl-2010-A Joint Rule Selection Model for Hierarchical Phrase-Based Translation
Author: Lei Cui ; Dongdong Zhang ; Mu Li ; Ming Zhou ; Tiejun Zhao
Abstract: In hierarchical phrase-based SMT systems, statistical models are integrated to guide the hierarchical rule selection for better translation performance. Previous work mainly focused on the selection of either the source side of a hierarchical rule or the target side of a hierarchical rule rather than considering both of them simultaneously. This paper presents a joint model to predict the selection of hierarchical rules. The proposed model is estimated based on four sub-models where the rich context knowledge from both source and target sides is leveraged. Our method can be easily incorporated into the practical SMT systems with the log-linear model framework. The experimental results show that our method can yield significant improvements in performance.
2 0.81222075 160 acl-2010-Learning Arguments and Supertypes of Semantic Relations Using Recursive Patterns
Author: Zornitsa Kozareva ; Eduard Hovy
Abstract: A challenging problem in open information extraction and text mining is the learning of the selectional restrictions of semantic relations. We propose a minimally supervised bootstrapping algorithm that uses a single seed and a recursive lexico-syntactic pattern to learn the arguments and the supertypes of a diverse set of semantic relations from the Web. We evaluate the performance of our algorithm on multiple semantic relations expressed using “verb”, “noun”, and “verb prep” lexico-syntactic patterns. Humanbased evaluation shows that the accuracy of the harvested information is about 90%. We also compare our results with existing knowledge base to outline the similarities and differences of the granularity and diversity of the harvested knowledge.
3 0.80551732 261 acl-2010-Wikipedia as Sense Inventory to Improve Diversity in Web Search Results
Author: Celina Santamaria ; Julio Gonzalo ; Javier Artiles
Abstract: Is it possible to use sense inventories to improve Web search results diversity for one word queries? To answer this question, we focus on two broad-coverage lexical resources of a different nature: WordNet, as a de-facto standard used in Word Sense Disambiguation experiments; and Wikipedia, as a large coverage, updated encyclopaedic resource which may have a better coverage of relevant senses in Web pages. Our results indicate that (i) Wikipedia has a much better coverage of search results, (ii) the distribution of senses in search results can be estimated using the internal graph structure of the Wikipedia and the relative number of visits received by each sense in Wikipedia, and (iii) associating Web pages to Wikipedia senses with simple and efficient algorithms, we can produce modified rankings that cover 70% more Wikipedia senses than the original search engine rankings. 1 Motivation The application of Word Sense Disambiguation (WSD) to Information Retrieval (IR) has been subject of a significant research effort in the recent past. The essential idea is that, by indexing and matching word senses (or even meanings) , the retrieval process could better handle polysemy and synonymy problems (Sanderson, 2000). In practice, however, there are two main difficulties: (i) for long queries, IR models implicitly perform disambiguation, and thus there is little room for improvement. This is the case with most standard IR benchmarks, such as TREC (trec.nist.gov) or CLEF (www.clef-campaign.org) ad-hoc collections; (ii) for very short queries, disambiguation j ul io @ l i uned . e s j avart s . @bec . uned . e s may not be possible or even desirable. This is often the case with one word and even two word queries in Web search engines. In Web search, there are at least three ways of coping with ambiguity: • • • Promoting diversity in the search results (Clarke negt al., 2008): given th seea query s”uolatssis”, the search engine may try to include representatives for different senses of the word (such as the Oasis band, the Organization for the Advancement of Structured Information Standards, the online fashion store, etc.) among the top results. Search engines are supposed to handle diversity as one of the multiple factors that influence the ranking. Presenting the results as a set of (labelled) cPlruessteenrtsi nragth tehre eth reansu as a a rsan ake sde lti ostf (Carpineto et al., 2009). Complementing search results with search suggestions (e.g. e”oaracshis band”, ”woitahsis s fashion store”) that serve to refine the query in the intended way (Anick, 2003). All of them rely on the ability of the search engine to cluster search results, detecting topic similarities. In all of them, disambiguation is implicit, a side effect of the process but not its explicit target. Clustering may detect that documents about the Oasis band and the Oasis fashion store deal with unrelated topics, but it may as well detect a group of documents discussing why one of the Oasis band members is leaving the band, and another group of documents about Oasis band lyrics; both are different aspects of the broad topic Oasis band. A perfect hierarchical clustering should distinguish between the different Oasis senses at a first level, and then discover different topics within each of the senses. Is it possible to use sense inventories to improve search results for one word queries? To answer 1357 Proce dingUsp opfs thaela 4, 8Stwhe Adnen u,a 1l1- M16e Jtiunlgy o 2f0 t1h0e. A ?c s 2o0c1ia0ti Aosnso focria Ctio nm fpourta Ctoiomnpault Laitniognuaislt Licisn,g puaigsetisc 1s357–136 , this question, we will focus on two broad-coverage lexical resources of a different nature: WordNet (Miller et al., 1990), as a de-facto standard used in Word Sense Disambiguation experiments and many other Natural Language Processing research fields; and Wikipedia (www.wikipedia.org), as a large coverage and updated encyclopedic resource which may have a better coverage of relevant senses in Web pages. Our hypothesis is that, under appropriate conditions, any of the above mechanisms (clustering, search suggestions, diversity) might benefit from an explicit disambiguation (classification of pages in the top search results) using a wide-coverage sense inventory. Our research is focused on four relevant aspects of the problem: 1. Coverage: Are Wikipedia/Wordnet senses representative of search results? Otherwise, trying to make a disambiguation in terms of a fixed sense inventory would be meaningless. 2. If the answer to (1) is positive, the reverse question is also interesting: can we estimate search results diversity using our sense inven- tories? 3. Sense frequencies: knowing sense frequencies in (search results) Web pages is crucial to have a usable sense inventory. Is it possible to estimate Web sense frequencies from currently available information? 4. Classification: The association of Web pages to word senses must be done with some unsupervised algorithm, because it is not possible to hand-tag training material for every possible query word. Can this classification be done accurately? Can it be effective to promote diversity in search results? In order to provide an initial answer to these questions, we have built a corpus consisting of 40 nouns and 100 Google search results per noun, manually annotated with the most appropriate Wordnet and Wikipedia senses. Section 2 describes how this corpus has been created, and in Section 3 we discuss WordNet and Wikipedia coverage of search results according to our testbed. As this initial results clearly discard Wordnet as a sense inventory for the task, the rest of the paper mainly focuses on Wikipedia. In Section 4 we estimate search results diversity from our testbed, finding that the use of Wikipedia could substantially improve diversity in the top results. In Section 5 we use the Wikipedia internal link structure and the number of visits per page to estimate relative frequencies for Wikipedia senses, obtaining an estimation which is highly correlated with actual data in our testbed. Finally, in Section 6 we discuss a few strategies to classify Web pages into word senses, and apply the best classifier to enhance diversity in search results. The paper concludes with a discussion of related work (Section 7) and an overall discussion of our results in Section 8. 2 Test Set 2.1 Set of Words The most crucial step in building our test set is choosing the set of words to be considered. We are looking for words which are susceptible to form a one-word query for a Web search engine, and therefore we should focus on nouns which are used to denote one or more named entities. At the same time we want to have some degree of comparability with previous research on Word Sense Disambiguation, which points to noun sets used in Senseval/SemEval evaluation campaigns1 . Our budget for corpus annotation was enough for two persons-month, which limited us to handle 40 nouns (usually enough to establish statistically significant differences between WSD algorithms, although obviously limited to reach solid figures about the general behaviour of words in the Web). With these arguments in mind, we decided to choose: (i) 15 nouns from the Senseval-3 lexical sample dataset, which have been previously employed by (Mihalcea, 2007) in a related experiment (see Section 7); (ii) 25 additional words which satisfy two conditions: they are all ambiguous, and they are all names for music bands in one of their senses (not necessarily the most salient). The Senseval set is: {argument, arm, atmosphere, bank, degree, difference, disc, irmm-, age, paper, party, performance, plan, shelter, sort, source}. The bands set is {amazon, apple, camel, cell, columbia, cream, foreigner, fox, genesis, jaguar, oasis, pioneer, police, puma, rainbow, shell, skin, sun, tesla, thunder, total, traffic, trapeze, triumph, yes}. Fpoerz e,a trchiu noun, we looked up all its possible senses in WordNet 3.0 and in Wikipedia (using 1http://senseval.org 1358 Table 1: Coverage of Search Results: Wikipedia vs. WordNet Wikiped#ia documents # senses WordNe#t documents Senseval setava2il4a2b/1le0/u0sedassign8e7d7 to (5 s9o%me) senseavai9la2b/5le2/usedassigne6d96 to (4 s6o%m)e sense # senses BaTnodtsa lset868420//21774421323558 ((5546%%))17780/3/9911529995 (2 (342%%)) Wikipedia disambiguation pages). Wikipedia has an average of 22 senses per noun (25.2 in the Bands set and 16. 1in the Senseval set), and Wordnet a much smaller figure, 4.5 (3. 12 for the Bands set and 6.13 for the Senseval set). For a conventional dictionary, a higher ambiguity might indicate an excess of granularity; for an encyclopaedic resource such as Wikipedia, however, it is just an indication of larger coverage. Wikipedia en- tries for camel which are not in WordNet, for instance, include the Apache Camel routing and mediation engine, the British rock band, the brand of cigarettes, the river in Cornwall, and the World World War I fighter biplane. 2.2 Set of Documents We retrieved the 150 first ranked documents for each noun, by submitting the nouns as queries to a Web search engine (Google). Then, for each document, we stored both the snippet (small description of the contents of retrieved document) and the whole HTML document. This collection of documents contain an implicit new inventory of senses, based on Web search, as documents retrieved by a noun query are associated with some sense of the noun. Given that every document in the top Web search results is supposed to be highly relevant for the query word, we assume a ”one sense per document” scenario, although we allow annotators to assign more than one sense per document. In general this assumption turned out to be correct except in a few exceptional cases (such as Wikipedia disambiguation pages): only nine docu- ments received more than one WordNet sense, and 44 (1. 1% of all annotated pages) received more than one Wikipedia sense. 2.3 Manual Annotation We implemented an annotation interface which stored all documents and a short description for every Wordnet and Wikipedia sense. The annotators had to decide, for every document, whether there was one or more appropriate senses in each of the dictionaries. They were instructed to provide annotations for 100 documents per name; if an URL in the list was corrupt or not available, it had to be discarded. We provided 150 documents per name to ensure that the figure of 100 usable documents per name could be reached without problems. Each judge provided annotations for the 4,000 documents in the final data set. In a second round, they met and discussed their independent annotations together, reaching a consensus judgement for every document. 3 Coverage of Web Search Results: Wikipedia vs Wordnet Table 1 shows how Wikipedia and Wordnet cover the senses in search results. We report each noun subset separately (Senseval and bands subsets) as well as aggregated figures. The most relevant fact is that, unsurprisingly, Wikipedia senses cover much more search results (56%) than Wordnet (32%). If we focus on the top ten results, in the bands subset (which should be more representative of plausible web queries) Wikipedia covers 68% of the top ten documents. This is an indication that it can indeed be useful for promoting diversity or help clustering search results: even if 32% of the top ten documents are not covered by Wikipedia, it is still a representative source of senses in the top search results. We have manually examined all documents in the top ten results that are not covered by Wikipedia: a majority of the missing senses consists of names of (generally not well-known) companies (45%) and products or services (26%); the other frequent type (12%) of non annotated doc- ument is disambiguation pages (from Wikipedia and also from other dictionaries). It is also interesting to examine the degree of overlap between Wikipedia and Wordnet senses. Being two different types of lexical resource, they might have some degree of complementarity. Table 2 shows, however, that this is not the case: most of the (annotated) documents either fit Wikipedia senses (26%) or both Wikipedia and Wordnet (29%), and just 3% fit Wordnet only. 1359 Table 2: Overlap between Wikipedia and Wordnet in Search Results # documents annotated with Senseval setWikipe60di7a ( &40 W%o)rdnetWi2k7ip0e (d1i8a% on)lyWo8r9d (n6e%t o)nly534no (3n6e%) BaTnodtsa slet1517729 ( (2239%%))1708566 (3 (216%%))12176 ( (13%%))11614195 ( (4415%%)) Therefore, Wikipedia seems to extend the coverage of Wordnet rather than providing complementary sense information. If we wanted to extend the coverage of Wikipedia, the best strategy seems to be to consider lists ofcompanies, products and services, rather than complementing Wikipedia with additional sense inventories. 4 Diversity in Google Search Results Once we know that Wikipedia senses are a representative subset of actual Web senses (covering more than half of the documents retrieved by the search engine), we can test how well search results respect diversity in terms of this subset of senses. Table 3 displays the number of different senses found at different depths in the search results rank, and the average proportion of total senses that they represent. These results suggest that diversity is not a major priority for ranking results: the top ten results only cover, in average, 3 Wikipedia senses (while the average number of senses listed in Wikipedia is 22). When considering the first 100 documents, this number grows up to 6.85 senses per noun. Another relevant figure is the frequency of the most frequent sense for each word: in average, 63% of the pages in search results belong to the most frequent sense of the query word. This is roughly comparable with most frequent sense figures in standard annotated corpora such as Semcor (Miller et al., 1993) and the Senseval/Semeval data sets, which suggests that diversity may not play a major role in the current Google ranking algorithm. Of course this result must be taken with care, because variability between words is high and unpredictable, and we are using only 40 nouns for our experiment. But what we have is a positive indication that Wikipedia could be used to improve diversity or cluster search results: potentially the first top ten results could cover 6.15 different senses in average (see Section 6.5), which would be a substantial growth. 5 Sense Frequency Estimators for Wikipedia Wikipedia disambiguation pages contain no systematic information about the relative importance of senses for a given word. Such information, however, is crucial in a lexicon, because sense distributions tend to be skewed, and knowing them can help disambiguation algorithms. We have attempted to use two estimators of expected sense distribution: • • Internal relevance of a word sense, measured as incoming alinnckes o ffo ar wthoer U seRnLs o, fm a given sense in Wikipedia. External relevance of a word sense, measured as ttheren naulm rebleevr aonfc vei osifts a f woro trhde s eUnRsLe, mofe a given sense (as reported in http://stats.grok.se). The number of internal incoming links is expected to be relatively stable for Wikipedia articles. As for the number of visits, we performed a comparison of the number of visits received by the bands noun subset in May, June and July 2009, finding a stable-enough scenario with one notorious exception: the number of visits to the noun Tesla raised dramatically in July, because July 10 was the anniversary of the birth of Nicola Tesla, and a special Google logo directed users to the Wikipedia page for the scientist. We have measured correlation between the relative frequencies derived from these two indicators and the actual relative frequencies in our testbed. Therefore, for each noun w and for each sense wi, we consider three values: (i) proportion of documents retrieved for w which are manually assigned to each sense wi; (ii) inlinks(wi) : relative amount of incoming links to each sense wi; and (iii) visits(wi) : relative number of visits to the URL for each sense wi. We have measured the correlation between these three values using a linear regression correlation coefficient, which gives a correlation value of .54 for the number of visits and of .71 for the number of incoming links. Both estimators seem 1360 Table 3: Diversity in Search Results according to Wikipedia F ir s t 12570 docsBave6n425.rd9854a6 s8get#snSe 65sien43. v68a3s27elarcthesTu6543l.o t5083as5lBvaen.r3d2a73s81gectovrSaegnso. 4f32v615aWlsiketpdaTs.3oe249tn01asle to be positively correlated with real relative frequencies in our testbed, with a strong preference for the number of links. We have experimented with weighted combinations of both indicators, using weights of the form (k, 1 k) , k ∈ {0, 0.1, 0.2 . . . 1}, reaching a maxi(mk,a1l c−okrre),lkati ∈on { 0of, .07.13, f0o.r2 t.h.e. following weights: − freq(wi) = 0.9∗inlinks(wi) +0. 1∗visits(wi) (1) This weighted estimator provides a slight advantage over the use of incoming links only (.73 vs .71). Overall, we have an estimator which has a strong correlation with the distribution of senses in our testbed. In the next section we will test its utility for disambiguation purposes. 6 Association of Wikipedia Senses to Web Pages We want to test whether the information provided by Wikipedia can be used to classify search results accurately. Note that we do not want to consider approaches that involve a manual creation of training material, because they can’t be used in practice. Given a Web page p returned by the search engine for the query w, and the set of senses w1 . . . wn listed in Wikipedia, the task is to assign the best candidate sense to p. We consider two different techniques: • A basic Information Retrieval approach, wAhe breas tche I dfoocrmumateionnts Ranetdr tvhael Wikipedia pages are represented using a Vector Space Model (VSM) and compared with a standard cosine measure. This is a basic approach which, if successful, can be used efficiently to classify search results. An approach based on a state-of-the-art supervised oWacShD b system, extracting training examples automatically from Wikipedia content. We also compute two baselines: • • • A random assignment of senses (precision is computed as itghnem ienvnter osfe oenfs tehse ( pnruemcibsieorn o isf senses, for every test case). A most frequent sense heuristic which uses our eosstitm fraetiqoune otf s sense frequencies acnhd u assigns the same sense (the most frequent) to all documents. Both are naive baselines, but it must be noted that the most frequent sense heuristic is usually hard to beat for unsupervised WSD algorithms in most standard data sets. We now describe each of the two main approaches in detail. 6.1 VSM Approach For each word sense, we represent its Wikipedia page in a (unigram) vector space model, assigning standard tf*idf weights to the words in the document. idf weights are computed in two different ways: 1. Experiment VSM computes inverse document frequencies in the collection of retrieved documents (for the word being considered). 2. Experiment VSM-GT uses the statistics provided by the Google Terabyte collection (Brants and Franz, 2006), i.e. it replaces the collection of documents with statistics from a representative snapshot of the Web. 3. Experiment VSM-mixed combines statistics from the collection and from the Google Terabyte collection, following (Chen et al., 2009). The document p is represented in the same vector space as the Wikipedia senses, and it is compared with each of the candidate senses wi via the cosine similarity metric (we have experimented 1361 with other similarity metrics such as χ2, but differences are irrelevant). The sense with the highest similarity to p is assigned to the document. In case of ties (which are rare), we pick the first sense in the Wikipedia disambiguation page (which in practice is like a random decision, because senses in disambiguation pages do not seem to be ordered according to any clear criteria). We have also tested a variant of this approach which uses the estimation of sense frequencies presented above: once the similarities are computed, we consider those cases where two or more senses have a similar score (in particular, all senses with a score greater or equal than 80% of the highest score). In that cases, instead of using the small similarity differences to select a sense, we pick up the one which has the largest frequency according to our estimator. We have applied this strategy to the best performing system, VSM-GT, resulting in experiment VSM-GT+freq. 6.2 WSD Approach We have used TiMBL (Daelemans et al., 2001), a state-of-the-art supervised WSD system which uses Memory-Based Learning. The key, in this case, is how to extract learning examples from the Wikipedia automatically. For each word sense, we basically have three sources of examples: (i) occurrences of the word in the Wikipedia page for the word sense; (ii) occurrences of the word in Wikipedia pages pointing to the page for the word sense; (iii) occurrences of the word in external pages linked in the Wikipedia page for the word sense. After an initial manual inspection, we decided to discard external pages for being too noisy, and we focused on the first two options. We tried three alternatives: • • • TiMBL-core uses only the examples found Tini MtheB page rfoer u tshees sense being atrmaipneleds. TiMBL-inlinks uses the examples found in Wikipedia pages pointing etxoa mthep sense being trained. TiMBL-all uses both sources of examples. In order to classify a page p with respect to the senses for a word w, we first disambiguate all occurrences of w in the page p. Then we choose the sense which appears most frequently in the page according to TiMBL results. In case of ties we pick up the first sense listed in the Wikipedia disambiguation page. We have also experimented with a variant of the approach that uses our estimation of sense frequencies, similarly to what we did with the VSM approach. In this case, (i) when there is a tie between two or more senses (which is much more likely than in the VSM approach), we pick up the sense with the highest frequency according to our estimator; and (ii) when no sense reaches 30% of the cases in the page to be disambiguated, we also resort to the most frequent sense heuristic (among the candidates for the page). This experiment is called TiMBL-core+freq (we discarded ”inlinks” and ”all” versions because they were clearly worse than ”core”). 6.3 Classification Results Table 4 shows classification results. The accuracy of systems is reported as precision, i.e. the number of pages correctly classified divided by the total number of predictions. This is approximately the same as recall (correctly classified pages divided by total number of pages) for our systems, because the algorithms provide an answer for every page containing text (actual coverage is 94% because some pages only contain text as part of an image file such as photographs and logotypes). Table 4: Classification Results Experiment Precision random most frequent sense (estimation) .19 .46 TiMBL-core TiMBL-inlinks TiMBL-all TiMBL-core+freq .60 .50 .58 .67 VSM VSM-GT VSM-mixed VSM-GT+freq .67 .68 .67 .69 All systems are significantly better than the random and most frequent sense baselines (using p < 0.05 for a standard t-test). Overall, both approaches (using TiMBL WSD machinery and using VSM) lead to similar results (.67 vs. .69), which would make VSM preferable because it is a simpler and more efficient approach. Taking a 1362 Figure 1: Precision/Coverage curves for VSM-GT+freq classification algorithm closer look at the results with TiMBL, there are a couple of interesting facts: • There is a substantial difference between using only examples itaalke dnif fferroemnc tehe b Wikipedia Web page for the sense being trained (TiMBL-core, .60) and using examples from the Wikipedia pages pointing to that page (TiMBL-inlinks, .50). Examples taken from related pages (even if the relationship is close as in this case) seem to be too noisy for the task. This result is compatible with findings in (Santamar ı´a et al., 2003) using the Open Directory Project to extract examples automatically. • Our estimation of sense frequencies turns oOuutr rto e tbiem very helpful sfeor f cases wcihesere t our TiMBL-based algorithm cannot provide an answer: precision rises from .60 (TiMBLcore) to .67 (TiMBL-core+freq). The difference is statistically significant (p < 0.05) according to the t-test. As for the experiments with VSM, the variations tested do not provide substantial improvements to the baseline (which is .67). Using idf frequencies obtained from the Google Terabyte corpus (instead of frequencies obtained from the set of retrieved documents) provides only a small improvement (VSM-GT, .68), and adding the estimation of sense frequencies gives another small improvement (.69). Comparing the baseline VSM with the optimal setting (VSM-GT+freq), the difference is small (.67 vs .69) but relatively robust (p = 0.066 according to the t-test). Remarkably, the use of frequency estimations is very helpful for the WSD approach but not for the SVM one, and they both end up with similar performance figures; this might indicate that using frequency estimations is only helpful up to certain precision ceiling. 6.4 Precision/Coverage Trade-off All the above experiments are done at maximal coverage, i.e., all systems assign a sense for every document in the test collection (at least for every document with textual content). But it is possible to enhance search results diversity without annotating every document (in fact, not every document can be assigned to a Wikipedia sense, as we have discussed in Section 3). Thus, it is useful to investigate which is the precision/coverage trade-off in our dataset. We have experimented with the best performing system (VSM-GT+freq), introducing a similarity threshold: assignment of a document to a sense is only done if the similarity of the document to the Wikipedia page for the sense exceeds the similarity threshold. We have computed precision and coverage for every threshold in the range [0.00 −0.90] (beyond 0e.v9e0ry coverage was null) anngde represented 0th] e(b breeysuolntds in Figure 1 (solid line). The graph shows that we 1363 can classify around 20% of the documents with a precision above .90, and around 60% of the documents with a precision of .80. Note that we are reporting disambiguation results using a conventional WSD test set, i.e., one in which every test case (every document) has been manually assigned to some Wikipedia sense. But in our Web Search scenario, 44% of the documents were not assigned to any Wikipedia sense: in practice, our classification algorithm would have to cope with all this noise as well. Figure 1 (dotted line) shows how the precision/coverage curve is affected when the algorithm attempts to disambiguate all documents retrieved by Google, whether they can in fact be assigned to a Wikipedia sense or not. At a coverage of 20%, precision drops approximately from .90 to .70, and at a coverage of 60% it drops from .80 to .50. We now address the question of whether this performance is good enough to improve search re- sults diversity in practice. 6.5 Using Classification to Promote Diversity We now want to estimate how the reported classification accuracy may perform in practice to enhance diversity in search results. In order to provide an initial answer to this question, we have re-ranked the documents for the 40 nouns in our testbed, using our best classifier (VSM-GT+freq) and making a list of the top-ten documents with the primary criterion of maximising the number of senses represented in the set, and the secondary criterion of maximising the similarity scores of the documents to their assigned senses. The algorithm proceeds as follows: we fill each position in the rank (starting at rank 1), with the document which has the highest similarity to some of the senses which are not yet represented in the rank; once all senses are represented, we start choosing a second representative for each sense, following the same criterion. The process goes on until the first ten documents are selected. We have also produced a number of alternative rankings for comparison purposes: clustering (centroids): this method applies eHriiengrarc (hciecnatlr Agglomerative Clustering which proved to be the most competitive clustering algorithm in a similar task (Artiles et al., 2009) to the set of search results, forcing the algorithm to create ten clusters. The centroid of each cluster is then selected Table 5: Enhancement of Search Results Diversity • – – rank@10 # senses coverage Original rank2.8049% Wikipedia 4.75 77% clustering (centroids) 2.50 42% clustering (top ranked) 2.80 46% random 2.45 43% upper bound6.1597% as one of the top ten documents in the new rank. • clustering (top ranked): Applies the same clustering algorithm, db u)t: tAhpisp lti emse t tehe s top ranked document (in the original Google rank) of each cluster is selected. • • random: Randomly selects ten documents frraonmd otmhe: :se Rt aofn dreomtrielyve sde lreecstuslts te. upper bound: This is the maximal diversity tuhpapt can o beu nodb:tai Tnheids iins our mteasxtbiemda. lN doivteer tshitayt coverage is not 100%, because some words have more than ten meanings in Wikipedia and we are only considering the top ten documents. All experiments have been applied on the full set of documents in the testbed, including those which could not be annotated with any Wikipedia sense. Coverage is computed as the ratio of senses that appear in the top ten results compared to the number of senses that appear in all search results. Results are presented in Table 5. Note that diversity in the top ten documents increases from an average of 2.80 Wikipedia senses represented in the original search engine rank, to 4.75 in the modified rank (being 6.15 the upper bound), with the coverage of senses going from 49% to 77%. With a simple VSM algorithm, the coverage of Wikipedia senses in the top ten results is 70% larger than in the original ranking. Using Wikipedia to enhance diversity seems to work much better than clustering: both strategies to select a representative from each cluster are unable to improve the diversity of the original ranking. Note, however, that our evaluation has a bias towards using Wikipedia, because only Wikipedia senses are considered to estimate diversity. Of course our results do not imply that the Wikipedia modified rank is better than the original 1364 Google rank: there are many other factors that influence the final ranking provided by a search engine. What our results indicate is that, with simple and efficient algorithms, Wikipedia can be used as a reference to improve search results diversity for one-word queries. 7 Related Work Web search results clustering and diversity in search results are topics that receive an increasing attention from the research community. Diversity is used both to represent sub-themes in a broad topic, or to consider alternative interpretations for ambiguous queries (Agrawal et al., 2009), which is our interest here. Standard IR test collections do not usually consider ambiguous queries, and are thus inappropriate to test systems that promote diversity (Sanderson, 2008); it is only recently that appropriate test collections are being built, such as (Paramita et al., 2009) for image search and (Artiles et al., 2009) for person name search. We see our testbed as complementary to these ones, and expect that it can contribute to foster research on search results diversity. To our knowledge, Wikipedia has not explicitly been used before to promote diversity in search results; but in (Gollapudi and Sharma, 2009), it is used as a gold standard to evaluate diversification algorithms: given a query with a Wikipedia disambiguation page, an algorithm is evaluated as promoting diversity when different documents in the search results are semantically similar to different Wikipedia pages (describing the alternative senses of the query). Although semantic similarity is measured automatically in this work, our results confirm that this evaluation strategy is sound, because Wikipedia senses are indeed representative of search results. (Clough et al., 2009) analyses query diversity in a Microsoft Live Search, using click entropy and query reformulation as diversity indicators. It was found that at least 9.5% - 16.2% of queries could benefit from diversification, although no correlation was found between the number of senses of a word in Wikipedia and the indicators used to discover diverse queries. This result does not discard, however, that queries where applying diversity is useful cannot benefit from Wikipedia as a sense inventory. In the context of clustering, (Carmel et al., 2009) successfully employ Wikipedia to enhance automatic cluster labeling, finding that Wikipedia labels agree with manual labels associated by humans to a cluster, much more than with signif- icant terms that are extracted directly from the text. In a similar line, both (Gabrilovich and Markovitch, 2007) and (Syed et al., 2008) provide evidence suggesting that categories of Wikipedia articles can successfully describe common concepts in documents. In the field of Natural Language Processing, there has been successful attempts to connect Wikipedia entries to Wordnet senses: (RuizCasado et al., 2005) reports an algorithm that provides an accuracy of 84%. (Mihalcea, 2007) uses internal Wikipedia hyperlinks to derive sensetagged examples. But instead of using Wikipedia directly as sense inventory, Mihalcea then manually maps Wikipedia senses into Wordnet senses (claiming that, at the time of writing the paper, Wikipedia did not consistently report ambiguity in disambiguation pages) and shows that a WSD system based on acquired sense-tagged examples reaches an accuracy well beyond an (informed) most frequent sense heuristic. 8 Conclusions We have investigated whether generic lexical resources can be used to promote diversity in Web search results for one-word, ambiguous queries. We have compared WordNet and Wikipedia and arrived to a number of conclusions: (i) unsurprisingly, Wikipedia has a much better coverage of senses in search results, and is therefore more appropriate for the task; (ii) the distribution of senses in search results can be estimated using the internal graph structure of the Wikipedia and the relative number of visits received by each sense in Wikipedia, and (iii) associating Web pages to Wikipedia senses with simple and efficient algorithms, we can produce modified rankings that cover 70% more Wikipedia senses than the original search engine rankings. We expect that the testbed created for this research will complement the - currently short - set of benchmarking test sets to explore search results diversity and query ambiguity. Our testbed is publicly available for research purposes at http://nlp.uned.es. Our results endorse further investigation on the use of Wikipedia to organize search results. Some limitations of our research, however, must be 1365 noted: (i) the nature of our testbed (with every search result manually annotated in terms of two sense inventories) makes it too small to extract solid conclusions on Web searches (ii) our work does not involve any study of diversity from the point of view of Web users (i.e. when a Web query addresses many different use needs in practice); research in (Clough et al., 2009) suggests that word ambiguity in Wikipedia might not be related with diversity of search needs; (iii) we have tested our classifiers with a simple re-ordering of search results to test how much diversity can be improved, but a search results ranking depends on many other factors, some of them more crucial than diversity; it remains to be tested how can we use document/Wikipedia associations to improve search results clustering (for instance, providing seeds for the clustering process) and to provide search suggestions. Acknowledgments This work has been partially funded by the Spanish Government (project INES/Text-Mess) and the Xunta de Galicia. References R. Agrawal, S. Gollapudi, A. Halverson, and S. Leong. 2009. Diversifying Search Results. In Proc. of WSDM’09. ACM. P. Anick. 2003. Using Terminological Feedback for Web Search Refinement : a Log-based Study. In Proc. ACM SIGIR 2003, pages 88–95. ACM New York, NY, USA. J. Artiles, J. Gonzalo, and S. Sekine. 2009. WePS 2 Evaluation Campaign: overview of the Web People Search Clustering Task. In 2nd Web People Search Evaluation Workshop (WePS 2009), 18th WWW Conference. 2009. T. Brants and A. Franz. 2006. Web 1T 5-gram, version 1. Philadelphia: Linguistic Data Consortium. D. Carmel, H. Roitman, and N. Zwerdling. 2009. Enhancing Cluster Labeling using Wikipedia. In Pro- ceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval, pages 139–146. ACM. C. Carpineto, S. Osinski, G. Romano, and Dawid Weiss. 2009. A Survey of Web Clustering Engines. ACM Computing Surveys, 41(3). Y. Chen, S. Yat Mei Lee, and C. Huang. 2009. PolyUHK: A Robust Information Extraction System for Web Personal Names. In Proc. WWW’09 (WePS2 Workshop). ACM. C. Clarke, M. Kolla, G. Cormack, O. Vechtomova, A. Ashkan, S. B ¨uttcher, and I. MacKinnon. 2008. Novelty and Diversity in Information Retrieval Evaluation. In Proc. SIGIR ’08, pages 659–666. ACM. P. Clough, M. Sanderson, M. Abouammoh, S. Navarro, and M. Paramita. 2009. Multiple Approaches to Analysing Query Diversity. In Proc. of SIGIR 2009. ACM. W. Daelemans, J. Zavrel, K. van der Sloot, and A. van den Bosch. 2001 . TiMBL: Tilburg Memory Based Learner, version 4.0, Reference Guide. Technical report, University of Antwerp. E. Gabrilovich and S. Markovitch. 2007. Computing Semantic Relatedness using Wikipedia-based Explicit Semantic Analysis. In Proceedings of The 20th International Joint Conference on Artificial Intelligence (IJCAI), Hyderabad, India. S. Gollapudi and A. Sharma. 2009. An Axiomatic Approach for Result Diversification. In Proc. WWW 2009, pages 381–390. ACM New York, NY, USA. R. Mihalcea. 2007. Using Wikipedia for Automatic Word Sense Disambiguation. In Proceedings of NAACL HLT, volume 2007. G. Miller, C. R. Beckwith, D. Fellbaum, Gross, and K. Miller. 1990. Wordnet: An on-line lexical database. International Journal of Lexicograph, 3(4). G.A Miller, C. Leacock, R. Tengi, and Bunker R. T. 1993. A Semantic Concordance. In Proceedings of the ARPA WorkShop on Human Language Technology. San Francisco, Morgan Kaufman. M. Paramita, M. Sanderson, and P. Clough. 2009. Diversity in Photo Retrieval: Overview of the ImageCLEFPhoto task 2009. CLEF working notes, 2009. M. Ruiz-Casado, E. Alfonseca, and P. Castells. 2005. Automatic Assignment of Wikipedia Encyclopaedic Entries to Wordnet Synsets. Advances in Web Intelligence, 3528:380–386. M. Sanderson. 2000. Retrieving with Good Sense. Information Retrieval, 2(1):49–69. M. Sanderson. 2008. Ambiguous Queries: Test Collections Need More Sense. In Proceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval, pages 499–506. ACM New York, NY, USA. C. Santamar ı´a, J. Gonzalo, and F. Verdejo. 2003. Automatic Association of Web Directories to Word Senses. Computational Linguistics, 29(3):485–502. Z. S. Syed, T. Finin, and Joshi. A. 2008. Wikipedia as an Ontology for Describing Documents. In Proc. ICWSM’08. 1366
4 0.72786576 169 acl-2010-Learning to Translate with Source and Target Syntax
Author: David Chiang
Abstract: Statistical translation models that try to capture the recursive structure of language have been widely adopted over the last few years. These models make use of varying amounts of information from linguistic theory: some use none at all, some use information about the grammar of the target language, some use information about the grammar of the source language. But progress has been slower on translation models that are able to learn the relationship between the grammars of both the source and target language. We discuss the reasons why this has been a challenge, review existing attempts to meet this challenge, and show how some old and new ideas can be combined into a sim- ple approach that uses both source and target syntax for significant improvements in translation accuracy.
5 0.72200823 237 acl-2010-Topic Models for Word Sense Disambiguation and Token-Based Idiom Detection
Author: Linlin Li ; Benjamin Roth ; Caroline Sporleder
Abstract: This paper presents a probabilistic model for sense disambiguation which chooses the best sense based on the conditional probability of sense paraphrases given a context. We use a topic model to decompose this conditional probability into two conditional probabilities with latent variables. We propose three different instantiations of the model for solving sense disambiguation problems with different degrees of resource availability. The proposed models are tested on three different tasks: coarse-grained word sense disambiguation, fine-grained word sense disambiguation, and detection of literal vs. nonliteral usages of potentially idiomatic expressions. In all three cases, we outper- form state-of-the-art systems either quantitatively or statistically significantly.
6 0.71501923 218 acl-2010-Structural Semantic Relatedness: A Knowledge-Based Method to Named Entity Disambiguation
7 0.71374893 172 acl-2010-Minimized Models and Grammar-Informed Initialization for Supertagging with Highly Ambiguous Lexicons
8 0.70936537 48 acl-2010-Better Filtration and Augmentation for Hierarchical Phrase-Based Translation Rules
9 0.70933372 53 acl-2010-Blocked Inference in Bayesian Tree Substitution Grammars
10 0.70902109 211 acl-2010-Simple, Accurate Parsing with an All-Fragments Grammar
11 0.70823967 114 acl-2010-Faster Parsing by Supertagger Adaptation
12 0.70606208 26 acl-2010-All Words Domain Adapted WSD: Finding a Middle Ground between Supervision and Unsupervision
13 0.70140064 71 acl-2010-Convolution Kernel over Packed Parse Forest
14 0.70128143 152 acl-2010-It Makes Sense: A Wide-Coverage Word Sense Disambiguation System for Free Text
15 0.70073724 113 acl-2010-Extraction and Approximation of Numerical Attributes from the Web
16 0.7003094 87 acl-2010-Discriminative Modeling of Extraction Sets for Machine Translation
17 0.69897366 76 acl-2010-Creating Robust Supervised Classifiers via Web-Scale N-Gram Data
18 0.69630039 156 acl-2010-Knowledge-Rich Word Sense Disambiguation Rivaling Supervised Systems
19 0.69562429 162 acl-2010-Learning Common Grammar from Multilingual Corpus
20 0.69261324 109 acl-2010-Experiments in Graph-Based Semi-Supervised Learning Methods for Class-Instance Acquisition