acl acl2013 acl2013-228 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Dan Goldwasser ; Dan Roth
Abstract: Semantic parsing is a domain-dependent process by nature, as its output is defined over a set of domain symbols. Motivated by the observation that interpretation can be decomposed into domain-dependent and independent components, we suggest a novel interpretation model, which augments a domain dependent model with abstract information that can be shared by multiple domains. Our experiments show that this type of information is useful and can reduce the annotation effort significantly when moving between domains.
Reference: text
sentIndex sentText sentNum sentScore
1 edu Abstract Semantic parsing is a domain-dependent process by nature, as its output is defined over a set of domain symbols. [sent-3, score-0.267]
2 Motivated by the observation that interpretation can be decomposed into domain-dependent and independent components, we suggest a novel interpretation model, which augments a domain dependent model with abstract information that can be shared by multiple domains. [sent-4, score-0.563]
3 1 Introduction Natural Language (NL) understanding can be intuitively understood as a general capacity, mapping words to entities and their relationships. [sent-6, score-0.081]
4 However, current work on automated NL understanding (typically referenced as semantic parsing (Zettlemoyer and Collins, 2005; Wong and Mooney, 2007; Chen and Mooney, 2008; Kwiatkowski et al. [sent-7, score-0.143]
5 , 2011)) is restricted to a given output domain1 (or task) consisting of a closed set of meaning representation symbols, describing domains such as robotic soccer, database queries and flight ordering systems. [sent-9, score-0.467]
6 In this work, we take a first step towards constructing a semantic interpreter that can leverage information from multiple tasks. [sent-10, score-0.102]
7 This is not a straight forward objective the domain specific nature of semantic interpretation, as described in the current literature, does not allow for an easy move between domains. [sent-11, score-0.217]
8 For example, a system trained for the task of understanding database queries will not be of any use when it will be given a sentence describing robotic soccer instructions. [sent-12, score-0.508]
9 In order to understand this difficulty, a closer – look at semantic parsing is required. [sent-13, score-0.143]
10 Given a sentence, the interpretation process breaks it into a 1The term domain is overloaded in NLP; in this work we use it to refer to the set of output symbols. [sent-14, score-0.367]
11 edu s set of interdependent decisions, which rely on an underlying representation mapping words to symbols and syntactic patterns into compositional decisions. [sent-16, score-0.295]
12 This representation takes into account domain specific information (e. [sent-17, score-0.15]
13 , a lexicon mapping phrases to a domain predicate) and is therefore of little use when moving to a different domain. [sent-19, score-0.266]
14 In this work, we attempt to develop a domain independent approach to semantic parsing. [sent-20, score-0.278]
15 We do it by developing a layer of representation that is applicable to multiple domains. [sent-21, score-0.159]
16 Specifically, we add an intermediate layer capturing shallow semantic relations between the input sentence constituents. [sent-22, score-0.419]
17 Unlike semantic parsing which maps the input to a closed set of symbols, this layer can be used to identify general predicate-argument structures in the input sentence. [sent-23, score-0.529]
18 The following example demonstrates the key idea behind our representation two sentences from two different domains have a similar intermediate structure. [sent-24, score-0.126]
19 , the word corresponding to a logical function is identified as a PRED). [sent-28, score-0.224]
20 Note that it does not use any domain specific information, for example, the PRED label assigned to the word “kicks ” indicates that this word is the predicate of the sentence, not a specific domain predicate (e. [sent-29, score-0.436]
21 The instepremciefidcia dtoe layer can c baete r (eeu. [sent-32, score-0.159]
22 The logical output associated with the second sentence is taken from a different domain, using a different set of output symbols, however it shares the same predicate-argument structure. [sent-36, score-0.306]
23 The mismatch between the domain independent (linguistic) structure and logical structures typically stems from technical considerations, as the domain logical language is designed according to an application-specific logic and not according to linguistic considerations. [sent-40, score-0.903]
24 2 Semantic Interpretation Model Our model consists of both domain-dependent (mapping between text and a closed set of symbols) and domain independent (abstract predicateargument structures) information. [sent-43, score-0.322]
25 We formulate the joint interpretation process as a structured prediction problem, mapping a NL input sentence (x), to its highest ranking interpretation and abstract structure (y). [sent-44, score-0.507]
26 The decision is quantified using a linear objective, which uses a vector w, mapping features to weights and a feature function Φ which maps the output decision to a feature vector. [sent-45, score-0.351]
27 The output interpretation y is described using a sub- set of first order logic, consisting of typed constants (e. [sent-46, score-0.312]
28 , robotic soccer player), functions capturing relations between entities, and their properties (e. [sent-48, score-0.515]
29 , pass(x, y), where pass is a function symbol and x, y are typed arguments). [sent-50, score-0.201]
30 We use data taken from two grounded domains, describing robotic soccer events and household situations. [sent-51, score-0.655]
31 , 2010) and formalize semantic inference as an Integer Linear Program (ILP). [sent-55, score-0.067]
32 1 Domain-Dependent Model Interpretation is composed of several decisions, capturing mapping of input tokens to logical fragments (first order) and their composition into larger fragments (second). [sent-60, score-0.472]
33 We encode a first-order decision as αcs, a binary variable indicating that constituent c is aligned with the logical symbol s. [sent-61, score-0.582]
34 A second-order decision βcs,dt, is encoded as a binary variable indicating that the symbol t (associated with constituent d) is an argument of a function s (associated with constituent c). [sent-62, score-0.411]
35 Φ1 depends on lexical information: each mapping of a lexical item c to a domain symbol s generates a feature. [sent-68, score-0.345]
36 In addition each combination of a lexical item c and an symbol type generates a feature. [sent-69, score-0.114]
37 Φ2 captures a pair of symbols and their alignment to lexical items. [sent-70, score-0.146]
38 Given a second-order decision βcs,dt, a feature is generated considering the normalized distance between the head words in the constituents c and d. [sent-71, score-0.143]
39 Another feature is generated for every composition of symbols (ignoring the alignment to the text). [sent-72, score-0.203]
40 2 Domain-Independent Information We enhance the decision process with information that abstracts over the attributes of specific domains by adding an intermediate layer consisting of the predicate-argument structure of the sentence. [sent-74, score-0.468]
41 Instead of relying on the mapping between Pink goalie and pink1, this model tries to identify an ARG using different means. [sent-76, score-0.196]
42 For example, the fact that it is preceded by a determiner, or capitalized provide useful cues. [sent-77, score-0.056]
43 We assume that these labels correspond to a binding to some logical symbol, and encode it as a constraint forcing the relations between the two models. [sent-80, score-0.38]
44 Moreover, since learning this layer is a by-product of the learning process (as it does not use any labeled data) forcing the connection between the decisions is the mechanism that drives learning this model. [sent-81, score-0.361]
45 Our domain-independent layer bears some similarity to other semantic tasks, most notably Semantic-Role Labeling (SRL) introduced in (Gildea and Jurafsky, 2002), in which identifying the predicate-argument structure is considered a preprocessing step, prior to assigning argument labels. [sent-82, score-0.343]
46 Unlike SRL, which aims to identify linguistic structures alone, in our framework these structures capture both natural-language and domain-language considerations. [sent-83, score-0.11]
47 Domain-Independent Decision Variables We add two new types of decisions abstracting over the domain-specific decisions. [sent-84, score-0.129]
48 The first (γ) captures local information helping to determine if a given constituent c is likely to have a label (i. [sent-86, score-0.059]
49 The second (δ) considers higher level structures, quantifying decisions over both the labels of the constituents c d as a predicate-argument pair. [sent-89, score-0.143]
50 , Model’s Features We use the following features: (1) Local Decisions Φ3 (γ(c)) use a feature indicating if c is capitalized, a set of features capturing the context of c (window of size 2), such as determiner and quantifier occurrences. [sent-91, score-0.158]
51 Finally we use a set of features capturing the suffix letters of c, these features are useful in identifying verb patterns. [sent-92, score-0.075]
52 (2) Global Decision Φ4(δ(c, d)) : a feature indicating the relative location of c compared to d in the input sentence. [sent-94, score-0.076]
53 Combined Model In order to consider both types of information we augment our decision model with the new variables, resulting in the following objective function (Eq. [sent-96, score-0.139]
54 In addition, we also add new constraints tying these new variables to semantic interpretation : ∀c ∈ x (γc → αc,s1 ∨ αc,s2 ∨ . [sent-99, score-0.297]
55 3 Learning the Combined Model The supervision to the learning process is given via data consisting of pairs of sentences and (domain specific) semantic interpretation. [sent-107, score-0.171]
56 Given that we have introduced additional variables that capture the more abstract predicate-argument structure of the text, we need to induce these as la- tent variables. [sent-108, score-0.093]
57 Our decision model maps an input sentence x, into a logical output y and predicateargument structure h. [sent-109, score-0.521]
58 We are only supplied with training data pertaining to the input (x) and output (y). [sent-110, score-0.076]
59 The dataset consists of triplets of the form - (x,u, y), where x is a NL sentence describing a situation (e. [sent-114, score-0.121]
60 , “He goes to the kitchen ”), u is a world state consisting of grounded relations (e. [sent-116, score-0.243]
61 , loc(John, Kitchen)) description, and y is a logical interpretation corresponding to x. [sent-118, score-0.4]
62 The original dataset was used for concept tagging, which does not include a compositional aspect. [sent-119, score-0.121]
63 We automatically generated the full logical structure by mapping the constants to function arguments. [sent-120, score-0.39]
64 We generated additional function symbols of the same relation, but of different arity when needed 3. [sent-121, score-0.146]
65 Our new dataset consists of 25 re- lation symbols (originally 15). [sent-122, score-0.199]
66 Robocup The Robocup dataset, originally introduced in (Chen and Mooney, 2008), describes robotic soccer events. [sent-124, score-0.44]
67 The dataset was collected for the purpose of constructing semantic parsers from ambiguous supervision and consists of both “noisy” and gold labeled data. [sent-125, score-0.245]
68 The noisy dataset 2Details omitted, see (Chang et al. [sent-126, score-0.135]
69 3For example, a unary relation symbol for “He plays”, and a binary for “He plays with a ball”. [sent-128, score-0.114]
70 was constructed by temporally aligning a stream of soccer events occurring during a robotic soccer match with human commentary describing the game. [sent-136, score-0.781]
71 This dataset consists of pairs (x, {y0, yk}), x aims a sentence aansedt {y0, yk} oisf a set so (fx events (logixc iasl formulas). [sent-137, score-0.145]
72 Onnde { oyf the}se i seav e snetts oisf eavsesunmts e(ldo gtocorrespond to the comment, however this is not guaranteed. [sent-138, score-0.056]
73 Semantic Interpretation Tasks We consider two of the tasks described in (Chen and Mooney, 2008) (1) Semantic Parsing requires generating the correct logical form given an input sentence. [sent-144, score-0.259]
74 (2) Matching, given a NL sentence and a set of several possible interpretation candidates, the system is required to identify the correct one. [sent-145, score-0.176]
75 We used the noisy Robocup dataset to initialize DOM-INIT, a noisy probabilistic model, constructed by taking statistics over the noisy robocup data and computing p(y|x). [sent-154, score-0.705]
76 o Grdi ienn x ies aligned etot every symbol i}n) every y yth wato irsd aligned sw ailthig nit. [sent-158, score-0.216]
77 e dTh teo probabilitQy of a matching (x, y)is computed as the product: Qin=1 p(yi |xi), where n is the number of symbolQs appearing in y, and xi, yi is the word 4In our model accuracy is equivalent to F-measure. [sent-159, score-0.059]
78 8462 Table 2: Results for the matching and parsing tasks. [sent-166, score-0.135]
79 Our system performs well on the matching task without any domain information. [sent-167, score-0.209]
80 Results for both parsing and matching tasks show that using domain-independent information improves results dramatically. [sent-168, score-0.135]
81 4 Knowledge Transfer Experiments We begin by studying the role of domainindependent information when very little domain information is available. [sent-171, score-0.212]
82 Domain-independent information is learned from the situated domain and domain-specific information (Robocup) available is the simple probabilistic model (DOM-INIT). [sent-172, score-0.262]
83 This model can be considered as a noisy probabilistic lexicon, without any domain-specific compositional information, which is only available through domain-independent information. [sent-173, score-0.15]
84 Most notably, performance for the matching task using only domain independent information (PRED-ARGS) was surprisingly good, with an accuracy of 0. [sent-175, score-0.27]
85 9, currently the highest for this task achieved without domain specific learning. [sent-178, score-0.15]
86 The second set of experiments study whether using domain independent information, when relevant (gold) domain-specific training data is available, improves learning. [sent-179, score-0.211]
87 domain (COMBINEDRL+S) and one relying on the Robocup training data alone (COMBINEDRL). [sent-182, score-0.15]
88 The results, summarized in table 3, consistently show that transferring domain independent information is helpful, and helps push the learned models beyond the supervision offered by the relevant domain training data. [sent-183, score-0.416]
89 Our final – system, trained over the entire dataset achieves a 465 System# trainingParsing C O M B IN E D R L + S (C O M B I N E D R L ) ful250g5ame0 . [sent-184, score-0.053]
90 It achieves similar results to (B¨ orschinger et al. [sent-193, score-0.112]
91 , 2011), the current state-of-the-art for the parsing task over this dataset. [sent-194, score-0.076]
92 5 Conclusions In this paper, we took a first step towards a new kind of generalization in semantic parsing: constructing a model that is able to generalize to a new domain defined over a different set of symbols. [sent-198, score-0.252]
93 Our approach adds an additional hidden layer to the semantic interpretation process, capturing shallow but domain-independent semantic information, which can be shared by different domains. [sent-199, score-0.544]
94 We describe two settings; in the first, where only noisy lexical-level domainspecific information is available, we observe that the model learned in the other domain can be used to make up for the missing compositional information. [sent-201, score-0.3]
95 For example, in the matching task, even when no domain information is available, identifying the abstract predicate argument structure provides sufficient discriminatory power to identify the correct event in over 69% of the times. [sent-202, score-0.393]
96 Learning to sportscast: a test of grounded language acquisition. [sent-244, score-0.077]
97 Generative alignment and semantic parsing for learning from ambiguous supervision. [sent-282, score-0.178]
98 Inducing probabilistic ccg grammars from logical form with higher-order unification. [sent-290, score-0.224]
99 Learning synchronous grammars for semantic parsing with lambda calculus. [sent-297, score-0.143]
100 Learning to map sentences to logical form: Structured classification with probabilistic categorial grammars. [sent-303, score-0.224]
wordName wordTfidf (topN-words)
[('robocup', 0.406), ('soccer', 0.237), ('logical', 0.224), ('robotic', 0.203), ('interpretation', 0.176), ('layer', 0.159), ('domain', 0.15), ('arg', 0.148), ('symbols', 0.146), ('pred', 0.145), ('mooney', 0.125), ('goldwasser', 0.125), ('kitchen', 0.117), ('combinedrl', 0.115), ('goalie', 0.115), ('symbol', 0.114), ('orschinger', 0.112), ('situated', 0.112), ('forcing', 0.107), ('clarke', 0.102), ('kicks', 0.101), ('pink', 0.101), ('decision', 0.095), ('decisions', 0.095), ('nl', 0.092), ('pass', 0.087), ('intermediate', 0.083), ('chen', 0.082), ('noisy', 0.082), ('mapping', 0.081), ('yk', 0.08), ('grounded', 0.077), ('parsing', 0.076), ('capturing', 0.075), ('describing', 0.068), ('predicate', 0.068), ('bordes', 0.068), ('compositional', 0.068), ('semantic', 0.067), ('closed', 0.063), ('domainindependent', 0.062), ('independent', 0.061), ('zettlemoyer', 0.06), ('constituent', 0.059), ('matching', 0.059), ('composition', 0.057), ('ilp', 0.056), ('capitalized', 0.056), ('oisf', 0.056), ('supervision', 0.055), ('structures', 0.055), ('variables', 0.054), ('dataset', 0.053), ('ball', 0.052), ('transferred', 0.05), ('kwiatkowski', 0.05), ('consisting', 0.049), ('encode', 0.049), ('predicateargument', 0.048), ('constituents', 0.048), ('fw', 0.047), ('argmax', 0.046), ('pc', 0.046), ('constants', 0.046), ('games', 0.045), ('augment', 0.044), ('domains', 0.043), ('argument', 0.043), ('determiner', 0.042), ('indicating', 0.041), ('considerations', 0.041), ('output', 0.041), ('srl', 0.039), ('structure', 0.039), ('maps', 0.039), ('cd', 0.039), ('cp', 0.039), ('cs', 0.037), ('wong', 0.037), ('events', 0.036), ('chang', 0.036), ('constructing', 0.035), ('notably', 0.035), ('il', 0.035), ('ambiguous', 0.035), ('moving', 0.035), ('input', 0.035), ('onnde', 0.034), ('bk', 0.034), ('lhe', 0.034), ('pwe', 0.034), ('household', 0.034), ('irsd', 0.034), ('abstracting', 0.034), ('ienn', 0.034), ('etot', 0.034), ('danr', 0.034), ('discriminatory', 0.034), ('ldo', 0.034), ('rme', 0.034)]
simIndex simValue paperId paperTitle
same-paper 1 0.99999976 228 acl-2013-Leveraging Domain-Independent Information in Semantic Parsing
Author: Dan Goldwasser ; Dan Roth
Abstract: Semantic parsing is a domain-dependent process by nature, as its output is defined over a set of domain symbols. Motivated by the observation that interpretation can be decomposed into domain-dependent and independent components, we suggest a novel interpretation model, which augments a domain dependent model with abstract information that can be shared by multiple domains. Our experiments show that this type of information is useful and can reduce the annotation effort significantly when moving between domains.
2 0.11434459 36 acl-2013-Adapting Discriminative Reranking to Grounded Language Learning
Author: Joohyun Kim ; Raymond Mooney
Abstract: We adapt discriminative reranking to improve the performance of grounded language acquisition, specifically the task of learning to follow navigation instructions from observation. Unlike conventional reranking used in syntactic and semantic parsing, gold-standard reference trees are not naturally available in a grounded setting. Instead, we show how the weak supervision of response feedback (e.g. successful task completion) can be used as an alternative, experimentally demonstrating that its performance is comparable to training on gold-standard parse trees.
3 0.10277712 312 acl-2013-Semantic Parsing as Machine Translation
Author: Jacob Andreas ; Andreas Vlachos ; Stephen Clark
Abstract: Semantic parsing is the problem of deriving a structured meaning representation from a natural language utterance. Here we approach it as a straightforward machine translation task, and demonstrate that standard machine translation components can be adapted into a semantic parser. In experiments on the multilingual GeoQuery corpus we find that our parser is competitive with the state of the art, and in some cases achieves higher accuracy than recently proposed purpose-built systems. These results support the use of machine translation methods as an informative baseline in semantic parsing evaluations, and suggest that research in semantic parsing could benefit from advances in machine translation.
4 0.095416225 98 acl-2013-Cross-lingual Transfer of Semantic Role Labeling Models
Author: Mikhail Kozhevnikov ; Ivan Titov
Abstract: Semantic Role Labeling (SRL) has become one of the standard tasks of natural language processing and proven useful as a source of information for a number of other applications. We address the problem of transferring an SRL model from one language to another using a shared feature representation. This approach is then evaluated on three language pairs, demonstrating competitive performance as compared to a state-of-the-art unsupervised SRL system and a cross-lingual annotation projection baseline. We also consider the contribution of different aspects of the feature representation to the performance of the model and discuss practical applicability of this method. 1 Background and Motivation Semantic role labeling has proven useful in many natural language processing tasks, such as question answering (Shen and Lapata, 2007; Kaisser and Webber, 2007), textual entailment (Sammons et al., 2009), machine translation (Wu and Fung, 2009; Liu and Gildea, 2010; Gao and Vogel, 2011) and dialogue systems (Basili et al., 2009; van der Plas et al., 2009). Multiple models have been designed to automatically predict semantic roles, and a considerable amount of data has been annotated to train these models, if only for a few more popular languages. As the annotation is costly, one would like to leverage existing resources to minimize the human effort required to construct a model for a new language. A number of approaches to the construction of semantic role labeling models for new languages have been proposed. On one end of the scale is unsupervised SRL, such as Grenager and Manning (2006), which requires some expert knowledge, but no labeled data. It clusters together arguments that should bear the same semantic role, but does not assign a particular role to each cluster. On the other end is annotating a new dataset from scratch. There are also intermediate options, which often make use of similarities between languages. This way, if an accurate model exists for one language, it should help simplify the construction of a model for another, related language. The approaches in this third group often use parallel data to bridge the gap between languages. Cross-lingual annotation projection systems (Pad o´ and Lapata, 2009), for example, propagate information directly via word alignment links. However, they are very sensitive to the quality of parallel data, as well as the accuracy of a sourcelanguage model on it. An alternative approach, known as cross-lingual model transfer, or cross-lingual model adaptation, consists of modifying a source-language model to make it directly applicable to a new language. This usually involves constructing a shared feature representation across the two languages. McDonald et al. (201 1) successfully apply this idea to the transfer of dependency parsers, using part-of- speech tags as the shared representation of words. A later extension of T ¨ackstr o¨m et al. (2012) enriches this representation with cross-lingual word clusters, considerably improving the performance. In the case of SRL, a shared representation that is purely syntactic is likely to be insufficient, since structures with different semantics may be realized by the same syntactic construct, for example “in August” vs “in Britain”. However with the help of recently introduced cross-lingual word represen1190 Proce dingsS o f ita h,e B 5u1lgsta Arinan,u Aaulg Musete 4ti-n9g 2 o0f1 t3h.e ? Ac s2s0o1ci3a Atiosnso fcoirat Cio nm foprut Caotimonpaulta Lti nognuails Lti cnsg,u piasgteics 1 90–120 , tations, such as the cross-lingual clustering mentioned above or cross-lingual distributed word representations of Klementiev et al. (2012), we may be able to transfer models of shallow semantics in a similar fashion. In this work we construct a shared feature representation for a pair of languages, employing crosslingual representations of syntactic and lexical information, train a semantic role labeling model on one language and apply it to the other one. This approach yields an SRL model for a new language at a very low cost, effectively requiring only a source language model and parallel data. We evaluate on five (directed) language pairs EN-ZH, ZH-EN, EN-CZ, CZ-EN and EN-FR, where EN, FR, CZ and ZH denote English, French, Czech and Chinese, respectively. The transferred model is compared against two baselines: an unsupervised SRL system and a model trained on the output of a cross-lingual annotation projection system. In the next section we will describe our setup, then in section 3 present the shared feature representation we use, discuss the evaluation data and other technical aspects in section 4, present the results and conclude with an overview of related work. – 2 Setup The purpose of the study is not to develop a yet another semantic role labeling system any existing SRL system can (after some modification) be used in this setup but to assess the practical applicability of cross-lingual model transfer to this – – problem, compare it against the alternatives and identify its strong/weak points depending on a particular setup. 2.1 Semantic Role Labeling Model We consider the dependency-based version of semantic role labeling as described in Haji cˇ et al. (2009) and transfer an SRL model from one language to another. We only consider verbal predicates and ignore the predicate disambiguation stage. We also assume that the predicate identification information is available in most languages it can be obtained using a relatively simple heuristic based on part-of-speech tags. The model performs argument identification and classification (Johansson and Nugues, 2008) separately in a pipeline first each candidate is classified as being or not being a head of an argument phrase with respect to the predicate in question and then each of the arguments is assigned a role from a given inventory. The model is factorized over arguments the decisions regarding the classification of different arguments are made in– – – dependently of each other. With respect to the use of syntactic annotation we consider two options: using an existing dependency parser for the target language and obtaining one by means of cross-lingual transfer (see section 4.2). Following McDonald et al. (201 1), we assume that a part-of-speech tagger is available for the target language. 2.2 SRL in the Low-resource Setting Several approaches have been proposed to obtain an SRL model for a new language with little or no manual annotation. Unsupervised SRL models (Lang and Lapata, 2010) cluster the arguments of predicates in a given corpus according to their semantic roles. The performance of such models can be impressive, especially for those languages where semantic roles correlate strongly with syntactic relation of the argument to its predicate. However, assigning meaningful role labels to the resulting clusters requires additional effort and the model’s parameters generally need some adjustment for every language. If the necessary resources are already available for a closely related language, they can be utilized to facilitate the construction of a model for the target language. This can be achieved either by means of cross-lingual annotation projection (Yarowsky et al., 2001) or by cross-lingual model transfer (Zeman and Resnik, 2008). This last approach is the one we are considering in this work, and the other two options are treated as baselines. The unsupervised model will be further referred to as UNSUP and the projection baseline as PROJ. 2.3 Evaluation Measures We use the F1 measure as a metric for the argument identification stage and accuracy as an aggregate measure of argument classification performance. When comparing to the unsupervised SRL system the clustering evaluation measures are used instead. These are purity and collocation 1191 N1Ximajx|Gj∩ Ci| CO =N1Xjmiax|Gj∩ Ci|, PU = where Ci is the set of arguments in the i-th induced cluster, Gj is the set of arguments in the jth gold cluster and N is the total number of arguments. We report the harmonic mean ofthe two (Lang and Lapata, 2011) and denote it F1c to avoid confusing it with the supervised metric. 3 Model Transfer The idea of this work is to abstract the model away from the particular source language and apply it to a new one. This setup requires that we use the same feature representation for both languages, for example part-of-speech tags and dependency relation labels should be from the same inventory. Some features are not applicable to certain lan- guages because the corresponding phenomena are absent in them. For example, consider a strongly inflected language and an analytic one. While the latter can usually convey the information encoded in the word form in the former one (number, gender, etc.), finding a shared feature representation for such information is non-trivial. In this study we will confine ourselves to those features that are applicable to all languages in question, namely: part-of-speech tags, syntactic dependency structures and representations of the word’s identity. 3.1 Lexical Information We train a model on one language and apply it to a different one. In order for this to work, the words of the two languages have to be mapped into a common feature space. It is also desirable that closely related words from both languages have similar representations in this space. Word mapping. The first option is simply to use the source language words as the shared representation. Here every source language word would have itself as its representation and every target word would map into a source word that corresponds to it. In other words, we supply the model with a gloss of the target sentence. The mapping (bilingual dictionary) we use is derived from a word-aligned parallel corpus, by identifying, for each word in the target language, the word in the source language it is most often aligned to. Cross-lingual clusters. There is no guarantee that each of the words in the evaluation data is present in our dictionary, nor that the corresponding source-language word is present in the training data, so the model would benefit from the ability to generalize over closely related words. This can, for example, be achieved by using cross-lingual word clusters induced in T ¨ackstr o¨m et al. (2012). We incorporate these clusters as features into our model. 3.2 Syntactic Information Part-of-speech Tags. We map part-of-speech tags into the universal tagset following Petrov et al. (2012). This may have a negative effect on the performance of a monolingual model, since most part-of-speech tagsets are more fine-grained than the universal POS tags considered here. For example Penn Treebank inventory contains 36 tags and the universal POS tagset only 12. Since the finergrained POS tags often reflect more languagespecific phenomena, however, they would only be useful for very closely related languages in the cross-lingual setting. The universal part-of-speech tags used in evaluation are derived from gold-standard annotation for all languages except French, where predicted ones had to be used instead. Dependency Structure. Another important aspect of syntactic information is the dependency structure. Most dependency relation inventories are language-specific, and finding a shared representation for them is a challenging problem. One could map dependency relations into a simplified form that would be shared between languages, as it is done for part-of-speech tags in Petrov et al. (2012). The extent to which this would be useful, however, depends on the similarity of syntactic-semantic in– terfaces of the languages in question. In this work we discard the dependency relation labels where the inventories do not match and only consider the unlabeled syntactic dependency graph. Some discrepancies, such as variations in attachment order, may be present even there, but this does not appear to be the case with the datasets we use for evaluation. If a target language is poor in resources, one can obtain a dependency parser for the target language by means of cross-lingual model transfer (Zeman and Resnik, 2008). We 1192 take this into account and evaluate both using the original dependency structures and the ones obtained by means of cross-lingual model transfer. 3.3 The Model The model we use is based on that of Bj ¨orkelund et al. (2009). It is comprised of a set of linear classifiers trained using Liblinear (Fan et al., 2008). The feature model was modified to accommodate the cross-lingual cluster features and the reranker component was not used. We do not model the interaction between different argument roles in the same predicate. While this has been found useful, in the cross-lingual setup one has to be careful with the assumptions made. For example, modeling the sequence of roles using a Markov chain (Thompson et al., 2003) may not work well in the present setting, especially between distant languages, as the order or arguments is not necessarily preserved. Most constraints that prove useful for SRL (Chang et al., 2007) also require customization when applied to a new language, and some rely on languagespecific resources, such as a valency lexicon. Taking into account the interaction between different arguments of a predicate is likely to improve the performance of the transferred model, but this is outside the scope of this work. 3.4 Feature Selection Compatibility of feature representations is necessary but not sufficient for successful model transfer. We have to make sure that the features we use are predictive of similar outcomes in the two languages as well. Depending on the pair of languages in question, different aspects of the feature representation will retain or lose their predictive power. We can be reasonably certain that the identity of an argument word is predictive of its semantic role in any language, but it might or might not be true of, for example, the word directly preceding the argument word. It is therefore important to pre- SCPDGylOespoSntreslTabunc1lra:obsFel-daitnguplrdoaeusntpagd-elronwfu-dcsopeyrnsd c.eylafguhtorsia mepgnrhs vent the model from capturing overly specific aspects of the source language, which we do by confining the model to first-order features. We also avoid feature selection, which, performed on the source language, is unlikely to help the model to better generalize to the target one. The experiments confirm that feature selection and the use of second-order features degrade the performance of the transferred model. 3.5 Feature Groups For each word, we use its part-of-speech tag, cross-lingual cluster id, word identity (glossed, when evaluating on the target language) and its dependency relation to its parent. Features associated with an argument word include the attributes of the predicate word, the argument word, its parent, siblings and children, and the words directly preceding and following it. Also included are the sequences of part-of-speech tags and dependency relations on the path between the predicate and the argument. Since we are also interested in the impact of different aspects of the feature representation, we divide the features into groups as summarized in table 1 and evaluate their respective contributions to the performance of the model. If a feature group is enabled the model has access to the corre– sponding source of information. For example, if only POS group is enabled, the model relies on the part-of-speech tags of the argument, the predicate and the words to the right and left of the argument word. If Synt is enabled too, it also uses the POS tags of the argument’s parent, children and siblings. Word order information constitutes an implicit group that is always available. It includes the Pos it ion feature, which indicates whether the argument is located to the left or to the right of the predicate, and allows the model to look up the attributes of the words directly preceding and following the argument word. The model we compare against the baselines uses all applicable feature groups (Deprel is only used in EN-CZ and CZ-EN experiments with original syntax). 4 Evaluation 4.1 Datasets and Preprocessing Evaluation of the cross-lingual model transfer requires a rather specific kind of dataset. Namely, the data in both languages has to be annotated 1193 with the same set of semantic roles following the same (or compatible) guidelines, which is seldom the case. We have identified three language pairs for which such resources are available: EnglishChinese, English-Czech and English-French. The evaluation datasets for English and Chinese are those from the CoNLL Shared Task 2009 (Haji ˇc et al., 2009) (henceforth CoNLL-ST). Their annotation in the CoNLL-ST is not identical, but the guidelines for “core” semantic roles are similar (Kingsbury et al., 2004), so we evaluate only on core roles here. The data for the second language pair is drawn from the Prague Czech-English Dependency Treebank 2.0 (Haji ˇc et al., 2012), which we converted to a format similar to that of CoNLL-ST1 . The original annotation uses the tectogrammatical representation (Haji ˇc, 2002) and an inventory of semantic roles (or functors), most of which are interpretable across various predicates. Also note that the syntactic anno- tation of English and Czech in PCEDT 2.0 is quite similar (to the extent permitted by the difference in the structure of the two languages) and we can use the dependency relations in our experiments. For English-French, the English CoNLL-ST dataset was used as a source and the model was evaluated on the manually annotated dataset from van der Plas et al. (201 1). The latter contains one thousand sentences from the French part ofthe Europarl (Koehn, 2005) corpus, annotated with semantic roles following an adapted version of PropBank (Palmer et al., 2005) guidelines. The authors perform annotation projection from English to French, using a joint model of syntax and semantics and employing heuristics for filtering. We use a model trained on the output of this projection system as one of the baselines. The evaluation dataset is relatively small in this case, so we perform the transfer only one-way, from English to French. The part-of-speech tags in all datasets were replaced with the universal POS tags of Petrov et al. (2012). For Czech, we have augmented the map- pings to account for the tags that were not present in the datasets from which the original mappings were derived. Namely, tag “t” is mapped to “VERB” and “Y” to “PRON”. We use parallel data to construct a bilingual dictionary used in word mapping, as well as in the projection baseline. For English-Czech – 1see http://www.ml4nlp.de/code-and-data/treex2conll and English-French, the data is drawn from Europarl (Koehn, 2005), for English-Chinese from MultiUN (Eisele and Chen, 2010). The word alignments were obtained using GIZA++ (Och and Ney, 2003) and the intersection heuristic. – 4.2 Syntactic Transfer In the low-resource setting, we cannot always rely on the availability of an accurate dependency parser for the target language. If one is not available, the natural solution would be to use crosslingual model transfer to obtain it. Unfortunately, the models presented in the previous work, such as Zeman and Resnik (2008), McDonald et al. (201 1) and T ¨ackstr o¨m et al. (2012), were not made available, so we reproduced the direct transfer algorithm of McDonald et al. (201 1), using Malt parser (Nivre, 2008) and the same set of features. We did not reimplement the projected transfer algorithm, however, and used the default training procedure instead of perceptron-based learning. The dependency structure thus obtained is, of course, only a rough approximation even a much more sophisticated algorithm may not perform well when transferring syntax between such languages as Czech and English, given the inherent difference in their structure. The scores are shown in table 2. We will henceforth refer to the syntactic annotations that were provided with the datasets as original, as opposed to the annotations obtained by means of syntactic transfer. – 4.3 Baselines Unsupervised Baseline: We are using a version of the unsupervised semantic role induction system of Titov and Klementiev (2012a) adapted to SetupUAS, % Table2:SyntaciE C ZcN HNt- rE ZaCFnN HZRsfer34 692567acuracy,unlabe dat- tachment score (percent). Note that in case of French we evaluate against the output of a supervised system, since manual annotation is not available for this dataset. This score does not reflect the true performance of syntactic transfer. 1194 the shared feature representation considered in order to make the scores comparable with those of the transfer model and, more importantly, to enable evaluation on transferred syntax. Note that the original system, tailored to a more expressive language-specific syntactic representation and equipped with heuristics to identify active/passive voice and other phenomena, achieves higher scores than those we report here. Projection Baseline: The projection baseline we use for English-Czech and English-Chinese is a straightforward one: we label the source side of a parallel corpus using the source-language model, then identify those verbs on the target side that are aligned to a predicate, mark them as predicates and propagate the argument roles in the same fashion. A model is then trained on the resulting training data and applied to the test set. For English-French we instead use the output of a fully featured projection model of van der Plas et al. (201 1), published in the CLASSiC project. 5 Results In order to ensure that the results are consistent, the test sets, except for the French one, were partitioned into five equal parts (of 5 to 10 thousand sentences each, depending on the dataset) and the evaluation performed separately on each one. All evaluation figures for English, Czech or Chinese below are the average values over the five subsets. In case of French, the evaluation dataset is too small to split it further, so instead we ran the evaluation five times on a randomly selected 80% sample of the evaluation data and averaged over those. In both cases the results are consistent over the subsets, the standard deviation does not exceed 0.5% for the transfer system and projection baseline and 1% for the unsupervised system. 5.1 Argument Identification We summarize the results in table 3. Argument identification is known to rely heavily on syntactic information, so it is unsurprising that it proves inaccurate when transferred syntax is used. Our simple projection baseline suffers from the same problem. Even with original syntactic information available, the performance of argument identification is moderate. Note that the model of (van der Plas et al., 2011), though relying on more expressive syntax, only outperforms the transferred system by 3% (F1) on this task. SetupSyntaxTRANSPROJ ZEC NH Z- EFCZNRHt r a n s 3462 1. 536 142 35. 4269 Table3EZ C:N H- CFEZANHZRrgumeon rt ig identf56 7ic13 a. t27903ion,21569t10ra. 3976nsferd model vs. projection baseline, F1. Most unsupervised SRL approaches assume that the argument identification is performed by some external means, for example heuristically (Lang and Lapata, 2011). Such heuristics or unsupervised approaches to argument identification (Abend et al., 2009) can also be used in the present setup. 5.2 Argument Classification In the following tables, TRANS column contains the results for the transferred system, UNSUP for the unsupervised baseline and PROJ for projection baseline. We highlight in bold the higher score where the difference exceeds twice the maximum of the standard deviation estimates of the two results. Table 4 presents the unsupervised evaluation results. Note that the unsupervised model performs as well as the transferred one or better where the – – SetupSyntaxTRANSUNSUP ZEC NH Z- EFCZNRHt r a n s 768 93648. 34627 6 5873. 1769 TableEZ C4NHZ:- FCEZANHZRrgumoe nr itg clasi78 fi94 3c. a25136tion,8 7 r9a4263n. 07 sferd model vs. unsupervised baseline in terms of the clustering metric F1c (see section 2.3). 1195 SetupSyntaxTRANSPROJ ZEC NH Z- EFCZNRHt r a n s 657 053. 1 36456419. 372 Table5EZ C:N H- CFEZANHZRrgumeon rt ig clasif657ic1936a. t170 ion,65 9t3804ra. 20847nsferd model vs. projection baseline, accuracy. original syntactic dependencies are available. In the more realistic scenario with transferred syn- tax, however, the transferred model proves more accurate. In table 5 we compare the transferred system with the projection baseline. It is easy to see that the scores vary strongly depending on the language pair, due to both the difference in the annotation scheme used and the degree of relatedness between the languages. The drop in performance when transferring the model to another language is large in every case, though, see table 6. SetupTargetSource Table6:MoCEZdHeNZ l- FECaZNRcH urac67 y53169o. 017nthes87 o25670u. r1245ceandtrge language using original syntax. The source language scores for English vary between language pairs because of the difference in syntactic annotation and role subset used. We also include the individual F1 scores for the top-10 most frequent labels for EN-CZ transfer with original syntax in table 7. The model provides meaningful predictions here, despite low overall accuracy. Most of the labels2 are self-explanatory: Patient (PAT), Actor (ACT), Time (TWHEN), Effect (EFF), Location (LOC), Manner (MANN), Addressee (ADDR), Extent (EXT). CPHR marks the 2http://ufal.mff.cuni.cz/∼toman/pcedt/en/functors.html LabelFreq.F1Re.Pr. recall and precision for the top-10 most frequent roles. nominal part of a complex predicate, as in “to have [a plan]CPHR”, and DIR3 indicates destination. 5.3 Additional Experiments We now evaluate the contribution of different aspects of the feature representation to the performance of the model. Table 8 contains the results for English-French. FeaturesOrigTrans ferent feature subsets, using original and transferred syntactic information. The fact that the model performs slightly better with transferred syntax may be explained by two factors. Firstly, as we already mentioned, the original syntactic annotation is also produced automatically. Secondly, in the model transfer setup it is more important how closely the syntacticsemantic interface on the target side resembles that on the source side than how well it matches the “true” structure of the target language, and in this respect a transferred dependency parser may have an advantage over one trained on target-language data. The high impact of the Glos s features here 1196 may be partly attributed to the fact that the mapping is derived from the same corpus as the evaluation data Europarl (Koehn, 2005) and partly by the similarity between English and French in terms of word order, usage of articles and prepositions. The moderate contribution of the crosslingual cluster features are likely due to the insufficient granularity of the clustering for this task. For more distant language pairs, the contributions of individual feature groups are less interpretable, so we only highlight a few observations. First of all, both EN-CZ and CZ-EN benefit noticeably from the use of the original syntactic annotation, including dependency relations, but not from the transferred syntax, most likely due to the low syntactic transfer performance. Both perform better when lexical information is available, although – – the improvement is not as significant as in the case of French only up to 5%. The situation with Chinese is somewhat complicated in that adding lexical information here fails to yield an improvement in terms of the metric considered. This is likely due to the fact that we consider only the core roles, which can usually be predicted with high accuracy based on syntactic information alone. – 6 Related Work Development of robust statistical models for core NLP tasks is a challenging problem, and adaptation of existing models to new languages presents a viable alternative to exhaustive annotation for each language. Although the models thus obtained are generally imperfect, they can be further refined for a particular language and domain using techniques such as active learning (Settles, 2010; Chen et al., 2011). Cross-lingual annotation projection (Yarowsky et al., 2001) approaches have been applied ex- tensively to a variety of tasks, including POS tagging (Xi and Hwa, 2005; Das and Petrov, 2011), morphology segmentation (Snyder and Barzilay, 2008), verb classification (Merlo et al., 2002), mention detection (Zitouni and Florian, 2008), LFG parsing (Wr o´blewska and Frank, 2009), information extraction (Kim et al., 2010), SRL (Pad o´ and Lapata, 2009; van der Plas et al., 2011; Annesi and Basili, 2010; Tonelli and Pianta, 2008), dependency parsing (Naseem et al., 2012; Ganchev et al., 2009; Smith and Eisner, 2009; Hwa et al., 2005) or temporal relation prediction (Spreyer and Frank, 2008). Interestingly, it has also been used to propagate morphosyntactic information between old and modern versions of the same language (Meyer, 2011). Cross-lingual model transfer methods (McDonald et al., 2011; Zeman and Resnik, 2008; Durrett et al., 2012; Søgaard, 2011; Lopez et al., 2008) have also been receiving much attention recently. The basic idea behind model transfer is similar to that of cross-lingual annotation projection, as we can see from the way parallel data is used in, for example, McDonald et al. (201 1). A crucial component of direct transfer approaches is the unified feature representation. There are at least two such representations of lexical information (Klementiev et al., 2012; T ¨ackstr o¨m et al., 2012), but both work on word level. This makes it hard to account for phenomena that are expressed differently in the languages considered, for example the syntactic function of a certain word may be indicated by a preposition, inflection or word order, depending on the language. Accurate representation of such information would require an extra level of abstraction (Haji ˇc, 2002). A side-effect ofusing adaptation methods is that we are forced to use the same annotation scheme for the task in question (SRL, in our case), which in turn simplifies the development of cross-lingual tools for downstream tasks. Such representations are also likely to be useful in machine translation. Unsupervised semantic role labeling methods (Lang and Lapata, 2010; Lang and Lapata, 2011; Titov and Klementiev, 2012a; Lorenzo and Cerisara, 2012) also constitute an alternative to cross-lingual model transfer. For an overview of of semi-supervised approaches we refer the reader to Titov and Klementiev (2012b). 7 Conclusion We have considered the cross-lingual model transfer approach as applied to the task of semantic role labeling and observed that for closely related languages it performs comparably to annotation projection approaches. It allows one to quickly construct an SRL model for a new language without manual annotation or language-specific heuristics, provided an accurate model is available for one of the related languages along with a certain amount of parallel data for the two languages. While an1197 notation projection approaches require sentenceand word-aligned parallel data and crucially depend on the accuracy of the syntactic parsing and SRL on the source side of the parallel corpus, cross-lingual model transfer can be performed using only a bilingual dictionary. Unsupervised SRL approaches have their advantages, in particular when no annotated data is available for any of the related languages and there is a syntactic parser available for the target one, but the annotation they produce is not always sufficient. In applications such as Information Retrieval it is preferable to have precise labels, rather than just clusters of arguments, for example. Also note that when applying cross-lingual model transfer in practice, one can improve upon the performance of the simplistic model we use for evaluation, for example by picking the features manually, taking into account the properties of the target language. Domain adaptation techniques can also be employed to adjust the model to the target language. Acknowledgments The authors would like to thank Alexandre Klementiev and Ryan McDonald for useful suggestions and T ¨ackstr o¨m et al. (2012) for sharing the cross-lingual word representations. This research is supported by the MMCI Cluster of Excellence. References Omri Abend, Roi Reichart, and Ari Rappoport. 2009. Unsupervised argument identification for semantic role labeling. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP, ACL ’09, pages 28–36, Stroudsburg, PA, USA. Association for Computational Linguistics. Paolo Annesi and Roberto Basili. 2010. Cross-lingual alignment of FrameNet annotations through hidden Markov models. In Proceedings of the 11th international conference on Computational Linguistics and Intelligent Text Processing, CICLing’ 10, pages 12– 25, Berlin, Heidelberg. Springer-Verlag. Roberto Basili, Diego De Cao, Danilo Croce, Bonaventura Coppola, and Alessandro Moschitti. 2009. Cross-language frame semantics transfer in bilingual corpora. In Alexander F. Gelbukh, editor, Proceedings of the 10th International Conference on Computational Linguistics and Intelligent Text Pro- cessing, pages 332–345. Anders Bj ¨orkelund, Love Hafdell, and Pierre Nugues. 2009. Multilingual semantic role labeling. In Proceedings of the Thirteenth Conference on Computational Natural Language Learning (CoNLL 2009): Shared Task, pages 43–48, Boulder, Colorado, June. Association for Computational Linguistics. Ming-Wei Chang, Lev Ratinov, and Dan Roth. 2007. Guiding semi-supervision with constraint-driven learning. In ACL. Chenhua Chen, Alexis Palmer, and Caroline Sporleder. 2011. Enhancing active learning for semantic role labeling via compressed dependency trees. In Proceedings of 5th International Joint Conference on Natural Language Processing, pages 183–191, Chiang Mai, Thailand, November. Asian Federation of Natural Language Processing. Dipanjan Das and Slav Petrov. 2011. Unsupervised part-of-speech tagging with bilingual graph-based projections. Proceedings of the Association for Computational Linguistics. Greg Durrett, Adam Pauls, and Dan Klein. 2012. Syntactic transfer using a bilingual lexicon. In Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 1–1 1, Jeju Island, Korea, July. Association for Computational Linguistics. Andreas Eisele and Yu Chen. 2010. MultiUN: A multilingual corpus from United Nation documents. In Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC’10). European Language Resources Association (ELRA). Rong-En Fan, Kai-Wei Chang, Cho-Jui Hsieh, XiangRui Wang, and Chih-Jen Lin. 2008. LIBLINEAR: A library for large linear classification. Journal of Machine Learning Research, 9: 1871–1874. Kuzman Ganchev, Jennifer Gillenwater, and Ben Taskar. 2009. Dependency grammar induction via bitext projection constraints. In Proceedings of the 47th Annual Meeting of the ACL, pages 369–377, Stroudsburg, PA, USA. Association for Computational Linguistics. Qin Gao and Stephan Vogel. 2011. Corpus expansion for statistical machine translation with semantic role label substitution rules. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 294–298, Portland, Oregon, USA. Trond Grenager and Christopher D. Manning. 2006. Unsupervised discovery of a statistical verb lexicon. In Proceedings of EMNLP. Jan Haji cˇ. 2002. Tectogrammatical representation: Towards a minimal transfer in machine translation. In Robert Frank, editor, Proceedings of the 6th International Workshop on Tree Adjoining Grammars 1198 and Related Frameworks (TAG+6), pages 216— 226, Venezia. Universita di Venezia. Jan Haji cˇ, Massimiliano Ciaramita, Richard Johansson, Daisuke Kawahara, Maria Ant o`nia Mart ı´, Llu ı´s M `arquez, Adam Meyers, Joakim Nivre, Sebastian Pad o´, Jan Sˇt eˇp a´nek, Pavel Stra nˇ a´k, Mihai Surdeanu, Nianwen Xue, and Yi Zhang. 2009. The CoNLL2009 shared task: Syntactic and semantic dependencies in multiple languages. In Proceedings of the Thirteenth Conference on Computational Natural Language Learning (CoNLL 2009): Shared Task, pages 1–18, Boulder, Colorado. Jan Haji cˇ, Eva Haji cˇov a´, Jarmila Panevov a´, Petr Sgall, Ond ˇrej Bojar, Silvie Cinkov´ a, Eva Fuˇ c ´ıkov a´, Marie Mikulov a´, Petr Pajas, Jan Popelka, Ji ˇr´ ı Semeck´ y, Jana Sˇindlerov a´, Jan Sˇt eˇp a´nek, Josef Toman, Zde nˇka Ure sˇov a´, and Zden eˇk Zˇabokrtsk y´. 2012. Announcing Prague Czech-English dependency treebank 2.0. In Nicoletta Calzolari (Conference Chair), Khalid Choukri, Thierry Declerck, Mehmet U gˇur Doˇ gan, Bente Maegaard, Joseph Mariani, Jan Odijk, and Stelios Piperidis, editors, Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’12), Istanbul, Turkey, May. European Language Resources Association (ELRA). Rebecca Hwa, Philip Resnik, Amy Weinberg, Clara Cabezas, and Okan Kolak. 2005. Bootstrapping parsers via syntactic projection across parallel text. Natural Language Engineering, 11(3):3 11–325. Richard Johansson and Pierre Nugues. 2008. Dependency-based semantic role labeling of PropBank. In Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, pages 69–78, Honolulu, Hawaii. Michael Kaisser and Bonnie Webber. 2007. Question answering based on semantic roles. In ACL Workshop on Deep Linguistic Processing. Seokhwan Kim, Minwoo Jeong, Jonghoon Lee, and Gary Geunbae Lee. 2010. A cross-lingual annotation projection approach for relation detection. In Proceedings of the 23rd International Conference on Computational Linguistics, COLING ’ 10, pages 564–571, Stroudsburg, PA, USA. Association for Computational Linguistics. Paul Kingsbury, Nianwen Xue, and Martha Palmer. 2004. Propbanking in parallel. In In Proceedings of the Workshop on the Amazing Utility of Parallel and Comparable Corpora, in conjunction with LREC’04. Alexandre Klementiev, Ivan Titov, and Binod Bhattarai. 2012. Inducing crosslingual distributed representations of words. In Proceedings of the International Conference on Computational Linguistics (COLING), Bombay, India. Philipp Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In Conference Proceedings: the tenth Machine Translation Summit, pages 79–86, Phuket, Thailand. AAMT. Joel Lang and Mirella Lapata. 2010. Unsupervised induction of semantic roles. In Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, pages 939–947, Los Angeles, California, June. Association for Computational Linguistics. Joel Lang and Mirella Lapata. 2011. Unsupervised semantic role induction via split-merge clustering. In Proc. of Annual Meeting of the Association for Computational Linguistics (ACL). Ding Liu and Daniel Gildea. 2010. Semantic role features for machine translation. In Proceedings of the 23rd International Conference on Computational Linguistics (Coling 2010), Beijing, China. Adam Lopez, Daniel Zeman, Michael Nossal, Philip Resnik, and Rebecca Hwa. 2008. Cross-language parser adaptation between related languages. In IJCNLP-08 Workshop on NLP for Less Privileged Languages, pages 35–42, Hyderabad, India, January. Alejandra Lorenzo and Christophe Cerisara. 2012. Unsupervised frame based semantic role induction: application to French and English. In Proceedings of the ACL 2012 Joint Workshop on Statistical Parsing and Semantic Processing of Morphologically Rich Languages, pages 30–35, Jeju, Republic of Korea, July. Association for Computational Linguistics. Ryan McDonald, Slav Petrov, and Keith Hall. 2011. Multi-source transfer of delexicalized dependency parsers. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, EMNLP ’ 11, pages 62–72, Stroudsburg, PA, USA. Association for Computational Linguistics. Paola Merlo, Suzanne Stevenson, Vivian Tsang, and Gianluca Allaria. 2002. A multi-lingual paradigm for automatic verb classification. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics (ACL’02), pages 207– 214, Philadelphia, PA. Roland Meyer. 2011. New wine in old wineskins?– Tagging old Russian via annotation projection from modern translations. Russian Linguistics, 35(2):267(15). Tahira Naseem, Regina Barzilay, and Amir Globerson. 2012. Selective sharing for multilingual dependency parsing. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics, pages 629–637, Jeju Island, Korea, July. Association for Computational Linguistics. Joakim Nivre. 2008. Algorithms for deterministic incremental dependency parsing. Comput. Linguist., 34(4):513–553, December. 1199 Franz Josef Och and Hermann Ney. 2003. A systematic comparison of various statistical alignment models. Computational Linguistics, 29(1). Sebastian Pad o´ and Mirella Lapata. 2009. Crosslingual annotation projection for semantic roles. Journal of Artificial Intelligence Research, 36:307– 340. Martha Palmer, Daniel Gildea, and Paul Kingsbury. 2005. The Proposition Bank: An annotated corpus of semantic roles. Computational Linguistics, 31:71–105. Slav Petrov, Dipanjan Das, and Ryan McDonald. 2012. A universal part-of-speech tagset. In Proceedings of LREC, May. Mark Sammons, Vinod Vydiswaran, Tim Vieira, Nikhil Johri, Ming wei Chang, Dan Goldwasser, Vivek Srikumar, Gourab Kundu, Yuancheng Tu, Kevin Small, Joshua Rule, Quang Do, and Dan Roth. 2009. Relation alignment for textual entailment recognition. In Text Analysis Conference (TAC). Burr Settles. 2010. Active learning literature survey. Computer Sciences Technical Report, 1648. Dan Shen and Mirella Lapata. 2007. Using semantic roles to improve question answering. In EMNLP. David A Smith and Jason Eisner. 2009. Parser adaptation and projection with quasi-synchronous grammar features. In Proceedings of the 2009 Confer- ence on Empirical Methods in Natural Language Processing, pages 822–831. Association for Computational Linguistics. Benjamin Snyder and Regina Barzilay. 2008. Crosslingual propagation for morphological analysis. In Proceedings of the 23rd national conference on Artificial intelligence. Anders Søgaard. 2011. Data point selection for crosslanguage adaptation of dependency parsers. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, volume 2 of HLT ’11, pages 682–686, Stroudsburg, PA, USA. Association for Computational Linguistics. Kathrin Spreyer and Anette Frank. 2008. Projectionbased acquisition of a temporal labeller. Proceedings of IJCNLP 2008. Oscar T¨ ackstr o¨m, Ryan McDonald, and Jakob Uszkoreit. 2012. Cross-lingual word clusters for direct transfer of linguistic structure. In Proc. of the Annual Meeting of the North American Association of Computational Linguistics (NAACL), pages 477– 487, Montr ´eal, Canada. Cynthia A. Thompson, Roger Levy, and Christopher D. Manning. 2003. A generative model for seman- tic role labeling. In Proceedings of the 14th European Conference on Machine Learning, ECML 2003, pages 397–408, Dubrovnik, Croatia. Ivan Titov and Alexandre Klementiev. 2012a. A Bayesian approach to unsupervised semantic role induction. In Proc. of European Chapter of the Association for Computational Linguistics (EACL). Ivan Titov and Alexandre Klementiev. 2012b. Semisupervised semantic role labeling: Approaching from an unsupervised perspective. In Proceedings of the International Conference on Computational Linguistics (COLING), Bombay, India, December. Sara Tonelli and Emanuele Pianta. 2008. Frame information transfer from English to Italian. In Proceedings of LREC 2008. Lonneke van der Plas, James Henderson, and Paola Merlo. 2009. Domain adaptation with artificial data for semantic parsing of speech. In Proc. 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, pages 125–128, Boulder, Colorado. Lonneke van der Plas, Paola Merlo, and James Henderson. 2011. Scaling up automatic cross-lingual semantic role annotation. In Proceedings of the 49th Annual Meeting of the Association for Computa- tional Linguistics: Human Language Technologies, HLT ’ 11, pages 299–304, Stroudsburg, PA, USA. Association for Computational Linguistics. Alina Wr o´blewska and Anette Frank. 2009. Crosslingual projection of LFG F-structures: Building an F-structure bank for Polish. In Eighth International Workshop on Treebanks and Linguistic Theories, page 209. Dekai Wu and Pascale Fung. 2009. Can semantic role labeling improve SMT? In Proceedings of 13th Annual Conference of the European Association for Machine Translation (EAMT 2009), Barcelona. Chenhai Xi and Rebecca Hwa. 2005. A backoff model for bootstrapping resources for non-English languages. In Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, pages 85 1–858, Stroudsburg, PA, USA. David Yarowsky, Grace Ngai, and Ricahrd Wicentowski. 2001. Inducing multilingual text analysis tools via robust projection across aligned corpora. In Proceedings of Human Language Technology Conference. Daniel Zeman and Philip Resnik. 2008. Crosslanguage parser adaptation between related lan- guages. In Proceedings of the IJCNLP-08 Workshop on NLP for Less Privileged Languages, pages 35– 42, Hyderabad, India, January. Asian Federation of Natural Language Processing. Imed Zitouni and Radu Florian. 2008. Mention detection crossing the language barrier. In Proceedings of the Conference on Empirical Methods in Natural Language Processing. 1200
5 0.094113976 212 acl-2013-Language-Independent Discriminative Parsing of Temporal Expressions
Author: Gabor Angeli ; Jakob Uszkoreit
Abstract: Temporal resolution systems are traditionally tuned to a particular language, requiring significant human effort to translate them to new languages. We present a language independent semantic parser for learning the interpretation of temporal phrases given only a corpus of utterances and the times they reference. We make use of a latent parse that encodes a language-flexible representation of time, and extract rich features over both the parse and associated temporal semantics. The parameters of the model are learned using a weakly supervised bootstrapping approach, without the need for manually tuned parameters or any other language expertise. We achieve state-of-the-art accuracy on all languages in the TempEval2 temporal normalization task, reporting a 4% improvement in both English and Spanish accuracy, and to our knowledge the first results for four other languages.
6 0.091259524 314 acl-2013-Semantic Roles for String to Tree Machine Translation
7 0.087030925 215 acl-2013-Large-scale Semantic Parsing via Schema Matching and Lexicon Extension
8 0.085102722 112 acl-2013-Dependency Parser Adaptation with Subtrees from Auto-Parsed Target Domain Data
9 0.082414187 291 acl-2013-Question Answering Using Enhanced Lexical Semantic Models
10 0.079131544 176 acl-2013-Grounded Unsupervised Semantic Parsing
11 0.077681534 388 acl-2013-Word Alignment Modeling with Context Dependent Deep Neural Network
12 0.074783385 306 acl-2013-SPred: Large-scale Harvesting of Semantic Predicates
13 0.071349971 347 acl-2013-The Role of Syntax in Vector Space Models of Compositional Semantics
14 0.070176698 80 acl-2013-Chinese Parsing Exploiting Characters
15 0.06927146 313 acl-2013-Semantic Parsing with Combinatory Categorial Grammars
16 0.069217294 275 acl-2013-Parsing with Compositional Vector Grammars
17 0.069024213 237 acl-2013-Margin-based Decomposed Amortized Inference
18 0.067820787 263 acl-2013-On the Predictability of Human Assessment: when Matrix Completion Meets NLP Evaluation
19 0.067775242 32 acl-2013-A relatedness benchmark to test the role of determiners in compositional distributional semantics
20 0.066743672 216 acl-2013-Large tagset labeling using Feed Forward Neural Networks. Case study on Romanian Language
topicId topicWeight
[(0, 0.202), (1, -0.01), (2, -0.047), (3, -0.063), (4, -0.058), (5, 0.036), (6, 0.031), (7, -0.005), (8, -0.025), (9, 0.039), (10, -0.012), (11, -0.03), (12, 0.029), (13, -0.049), (14, -0.008), (15, -0.012), (16, 0.017), (17, 0.01), (18, 0.015), (19, -0.044), (20, -0.016), (21, 0.009), (22, -0.065), (23, 0.039), (24, 0.025), (25, 0.016), (26, 0.016), (27, 0.046), (28, 0.018), (29, 0.04), (30, -0.001), (31, 0.039), (32, 0.043), (33, 0.025), (34, 0.046), (35, -0.005), (36, -0.011), (37, -0.038), (38, 0.002), (39, 0.146), (40, -0.017), (41, 0.042), (42, -0.016), (43, -0.085), (44, -0.004), (45, 0.13), (46, 0.076), (47, -0.043), (48, -0.025), (49, 0.026)]
simIndex simValue paperId paperTitle
same-paper 1 0.92213041 228 acl-2013-Leveraging Domain-Independent Information in Semantic Parsing
Author: Dan Goldwasser ; Dan Roth
Abstract: Semantic parsing is a domain-dependent process by nature, as its output is defined over a set of domain symbols. Motivated by the observation that interpretation can be decomposed into domain-dependent and independent components, we suggest a novel interpretation model, which augments a domain dependent model with abstract information that can be shared by multiple domains. Our experiments show that this type of information is useful and can reduce the annotation effort significantly when moving between domains.
2 0.73946756 215 acl-2013-Large-scale Semantic Parsing via Schema Matching and Lexicon Extension
Author: Qingqing Cai ; Alexander Yates
Abstract: Supervised training procedures for semantic parsers produce high-quality semantic parsers, but they have difficulty scaling to large databases because of the sheer number of logical constants for which they must see labeled training data. We present a technique for developing semantic parsers for large databases based on a reduction to standard supervised training algorithms, schema matching, and pattern learning. Leveraging techniques from each of these areas, we develop a semantic parser for Freebase that is capable of parsing questions with an F1 that improves by 0.42 over a purely-supervised learning algorithm.
3 0.72138381 313 acl-2013-Semantic Parsing with Combinatory Categorial Grammars
Author: Yoav Artzi ; Nicholas FitzGerald ; Luke Zettlemoyer
Abstract: unkown-abstract
4 0.70182693 36 acl-2013-Adapting Discriminative Reranking to Grounded Language Learning
Author: Joohyun Kim ; Raymond Mooney
Abstract: We adapt discriminative reranking to improve the performance of grounded language acquisition, specifically the task of learning to follow navigation instructions from observation. Unlike conventional reranking used in syntactic and semantic parsing, gold-standard reference trees are not naturally available in a grounded setting. Instead, we show how the weak supervision of response feedback (e.g. successful task completion) can be used as an alternative, experimentally demonstrating that its performance is comparable to training on gold-standard parse trees.
5 0.68936247 176 acl-2013-Grounded Unsupervised Semantic Parsing
Author: Hoifung Poon
Abstract: We present the first unsupervised approach for semantic parsing that rivals the accuracy of supervised approaches in translating natural-language questions to database queries. Our GUSP system produces a semantic parse by annotating the dependency-tree nodes and edges with latent states, and learns a probabilistic grammar using EM. To compensate for the lack of example annotations or question-answer pairs, GUSP adopts a novel grounded-learning approach to leverage database for indirect supervision. On the challenging ATIS dataset, GUSP attained an accuracy of 84%, effectively tying with the best published results by supervised approaches.
6 0.6538586 324 acl-2013-Smatch: an Evaluation Metric for Semantic Feature Structures
7 0.63992453 163 acl-2013-From Natural Language Specifications to Program Input Parsers
8 0.63797939 175 acl-2013-Grounded Language Learning from Video Described with Sentences
9 0.59958774 349 acl-2013-The mathematics of language learning
10 0.59555203 311 acl-2013-Semantic Neighborhoods as Hypergraphs
11 0.59444183 312 acl-2013-Semantic Parsing as Machine Translation
12 0.59389728 347 acl-2013-The Role of Syntax in Vector Space Models of Compositional Semantics
13 0.58623779 275 acl-2013-Parsing with Compositional Vector Grammars
14 0.58525443 382 acl-2013-Variational Inference for Structured NLP Models
15 0.57799488 1 acl-2013-"Let Everything Turn Well in Your Wife": Generation of Adult Humor Using Lexical Constraints
16 0.57514888 189 acl-2013-ImpAr: A Deterministic Algorithm for Implicit Semantic Role Labelling
17 0.57069504 14 acl-2013-A Novel Classifier Based on Quantum Computation
18 0.56855941 90 acl-2013-Conditional Random Fields for Responsive Surface Realisation using Global Features
19 0.56185645 161 acl-2013-Fluid Construction Grammar for Historical and Evolutionary Linguistics
20 0.55889595 21 acl-2013-A Statistical NLG Framework for Aggregated Planning and Realization
topicId topicWeight
[(0, 0.043), (6, 0.052), (11, 0.077), (24, 0.051), (26, 0.03), (35, 0.097), (42, 0.067), (48, 0.048), (64, 0.279), (70, 0.043), (88, 0.038), (90, 0.03), (95, 0.071)]
simIndex simValue paperId paperTitle
1 0.90931481 268 acl-2013-PATHS: A System for Accessing Cultural Heritage Collections
Author: Eneko Agirre ; Nikolaos Aletras ; Paul Clough ; Samuel Fernando ; Paula Goodale ; Mark Hall ; Aitor Soroa ; Mark Stevenson
Abstract: This paper describes a system for navigating large collections of information about cultural heritage which is applied to Europeana, the European Library. Europeana contains over 20 million artefacts with meta-data in a wide range of European languages. The system currently provides access to Europeana content with meta-data in English and Spanish. The paper describes how Natural Language Processing is used to enrich and organise this meta-data to assist navigation through Europeana and shows how this information is used within the system.
2 0.84694552 15 acl-2013-A Novel Graph-based Compact Representation of Word Alignment
Author: Qun Liu ; Zhaopeng Tu ; Shouxun Lin
Abstract: In this paper, we propose a novel compact representation called weighted bipartite hypergraph to exploit the fertility model, which plays a critical role in word alignment. However, estimating the probabilities of rules extracted from hypergraphs is an NP-complete problem, which is computationally infeasible. Therefore, we propose a divide-and-conquer strategy by decomposing a hypergraph into a set of independent subhypergraphs. The experiments show that our approach outperforms both 1-best and n-best alignments.
same-paper 3 0.81218737 228 acl-2013-Leveraging Domain-Independent Information in Semantic Parsing
Author: Dan Goldwasser ; Dan Roth
Abstract: Semantic parsing is a domain-dependent process by nature, as its output is defined over a set of domain symbols. Motivated by the observation that interpretation can be decomposed into domain-dependent and independent components, we suggest a novel interpretation model, which augments a domain dependent model with abstract information that can be shared by multiple domains. Our experiments show that this type of information is useful and can reduce the annotation effort significantly when moving between domains.
Author: Guido Boella ; Luigi Di Caro
Abstract: In this paper we present a technique to reveal definitions and hypernym relations from text. Instead of using pattern matching methods that rely on lexico-syntactic patterns, we propose a technique which only uses syntactic dependencies between terms extracted with a syntactic parser. The assumption is that syntactic information are more robust than patterns when coping with length and complexity of the sentences. Afterwards, we transform such syntactic contexts in abstract representations, that are then fed into a Support Vector Machine classifier. The results on an annotated dataset of definitional sentences demonstrate the validity of our approach overtaking current state-of-the-art techniques.
5 0.7939046 355 acl-2013-TransDoop: A Map-Reduce based Crowdsourced Translation for Complex Domain
Author: Anoop Kunchukuttan ; Rajen Chatterjee ; Shourya Roy ; Abhijit Mishra ; Pushpak Bhattacharyya
Abstract: Large amount of parallel corpora is required for building Statistical Machine Translation (SMT) systems. We describe the TransDoop system for gathering translations to create parallel corpora from online crowd workforce who have familiarity with multiple languages but are not expert translators. Our system uses a Map-Reduce-like approach to translation crowdsourcing where sentence translation is decomposed into the following smaller tasks: (a) translation ofconstituent phrases of the sentence; (b) validation of quality of the phrase translations; and (c) composition of complete sentence translations from phrase translations. Trans- Doop incorporates quality control mechanisms and easy-to-use worker user interfaces designed to address issues with translation crowdsourcing. We have evaluated the crowd’s output using the METEOR metric. For a complex domain like judicial proceedings, the higher scores obtained by the map-reduce based approach compared to complete sentence translation establishes the efficacy of our work.
6 0.7736823 22 acl-2013-A Structured Distributional Semantic Model for Event Co-reference
7 0.65702045 6 acl-2013-A Java Framework for Multilingual Definition and Hypernym Extraction
8 0.64407265 265 acl-2013-Outsourcing FrameNet to the Crowd
9 0.62171847 158 acl-2013-Feature-Based Selection of Dependency Paths in Ad Hoc Information Retrieval
10 0.60266268 176 acl-2013-Grounded Unsupervised Semantic Parsing
11 0.60195744 17 acl-2013-A Random Walk Approach to Selectional Preferences Based on Preference Ranking and Propagation
12 0.59779984 159 acl-2013-Filling Knowledge Base Gaps for Distant Supervision of Relation Extraction
13 0.59314609 99 acl-2013-Crowd Prefers the Middle Path: A New IAA Metric for Crowdsourcing Reveals Turker Biases in Query Segmentation
14 0.59209597 83 acl-2013-Collective Annotation of Linguistic Resources: Basic Principles and a Formal Model
15 0.58303553 215 acl-2013-Large-scale Semantic Parsing via Schema Matching and Lexicon Extension
16 0.58135772 36 acl-2013-Adapting Discriminative Reranking to Grounded Language Learning
17 0.57077545 224 acl-2013-Learning to Extract International Relations from Political Context
18 0.56028891 266 acl-2013-PAL: A Chatterbot System for Answering Domain-specific Questions
19 0.56026983 185 acl-2013-Identifying Bad Semantic Neighbors for Improving Distributional Thesauri
20 0.56016457 172 acl-2013-Graph-based Local Coherence Modeling