acl acl2013 acl2013-314 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Marzieh Bazrafshan ; Daniel Gildea
Abstract: We experiment with adding semantic role information to a string-to-tree machine translation system based on the rule extraction procedure of Galley et al. (2004). We compare methods based on augmenting the set of nonterminals by adding semantic role labels, and altering the rule extraction process to produce a separate set of rules for each predicate that encompass its entire predicate-argument structure. Our results demonstrate that the second approach is effective in increasing the quality of translations.
Reference: text
sentIndex sentText sentNum sentScore
1 We compare methods based on augmenting the set of nonterminals by adding semantic role labels, and altering the rule extraction process to produce a separate set of rules for each predicate that encompass its entire predicate-argument structure. [sent-3, score-1.462]
2 1 Introduction Statistical machine translation (SMT) has made considerable advances in using syntactic proper- ties of languages in both the training and the decoding of translation systems. [sent-5, score-0.42]
3 Over the past few years, many researchers have started to realize that incorporating semantic features of languages can also be effective in increasing the quality of translations, as they can model relationships that often are not derivable from syntactic structures. [sent-6, score-0.307]
4 Wu and Fung (2009) demonstrated the promise of using features based on semantic predicateargument structure in machine translation, using these feature to re-rank machine translation output. [sent-7, score-0.493]
5 In general, re-ranking approaches are limited by the set of translation hypotheses, leading to a desire to incorporate semantic features into the translation model used during MT decoding. [sent-8, score-0.58]
6 Liu and Gildea (2010) introduced two types of semantic features for tree-to-string machine translation. [sent-9, score-0.228]
7 These features model the reorderings and deletions of the semantic roles in the source sentence during decoding. [sent-10, score-0.438]
8 They showed that addition of these semantic features helps improve the quality of translations. [sent-11, score-0.228]
9 Since tree-to-string systems are trained on parse trees, they are constrained by the tree structures and are generally outperformed by string-to-tree systems. [sent-12, score-0.087]
10 (2012) integrated two discriminative feature-based models into a phrase-based SMT system, which used the semantic predicateargument structure of the source language. [sent-14, score-0.317]
11 Their first model defined features based on the context of a verbal predicate, to predict the target translation for that verb. [sent-15, score-0.232]
12 Their second model predicted the reordering direction between a predicate and its arguments from the source to the target sentence. [sent-16, score-0.555]
13 (2010) use a head-driven phrase structure grammar (HPSG) parser to add semantic representations to their translation rules. [sent-18, score-0.447]
14 In this paper, we use semantic role labels to enrich a string-to-tree translation system, and show that this approach can increase the BLEU (Papineni et al. [sent-19, score-0.745]
15 , 2004) translation rules from training data where the target side has been parsed and labeled with semantic roles. [sent-22, score-0.791]
16 Our general method of adding information to the syntactic tree is similar to the “tree grafting” ap- proach of Baker et al. [sent-23, score-0.184]
17 We modify the rule extraction procedure of Galley et al. [sent-25, score-0.192]
18 (2004) to produce rules representing the overall predicateargument structure of each verb, allowing us to model alternations in the mapping from syntax to semantics of the type described by Levin (1993). [sent-26, score-0.411]
19 1 Semantic Role Labeling Semantic Role Labeling (SRL) is the task of identifying the arguments of the predicates in a sentence, and classifying them into different argument labels. [sent-28, score-0.248]
20 Semantic roles can provide a level 419 Proce dingSsof oifa, th Beu 5l1gsarti Aan,An u aglu Mste 4e-ti9n2g 0 o1f3 t. [sent-29, score-0.21]
21 ” and “The door opened”, the word door has two different syntac- tic roles but only one semantic role in the two sentences. [sent-33, score-0.794]
22 Semantic arguments can be classified into core and non-core arguments (Palmer et al. [sent-34, score-0.417]
23 Non-core arguments add more information about the predicate but are not essential. [sent-37, score-0.542]
24 Automatic semantic role labelers have been developed by training classifiers on hand annotated data (Gildea and Jurafsky, 2000; Srikumar and Roth, 2011; Toutanova et al. [sent-38, score-0.518]
25 State-of-the-art semantic role labelers can predict the labels with accuracies of around 90%. [sent-40, score-0.641]
26 (2004) using the parses produced by the splitmerge parser of Petrov et al. [sent-43, score-0.038]
27 (2010), the refined nonterminals produced by the split-merge method can aid machine translation. [sent-46, score-0.202]
28 Furthermore, in all of our experiments, we exclude unary rules during extraction by ensuring that no rules will have the same span in the source side (Chung et al. [sent-47, score-0.637]
29 4 Semantically Enriched Rules (Method 1) In this method, we tag the target trees in the training corpus with semantic role labels, and extract the translation rules from the tagged corpus. [sent-53, score-1.036]
30 “Lending” is the predicate, “everybody” is argument 0, and “a hand” is argument 1 for the predicate. [sent-55, score-0.13]
31 S-8 NP-7-ARG11 victimized by NP-7-ARG02 NP-7-ARG11 受 NP-7-ARG02 Figure 2: A complete semantic rule. [sent-56, score-0.296]
32 We only label the core arguments of each predicate, to make sure that the rules are not too specific to the training data. [sent-58, score-0.543]
33 We attach each semantic label to the root of the subtree that it is labeling. [sent-59, score-0.319]
34 Figure 1 shows an example target tree after attaching the semantic roles. [sent-60, score-0.371]
35 We then run a GHKM rule extractor on the labeled training corpus and use the semantically enriched rules with a syntax-based decoder. [sent-61, score-0.544]
36 5 Complete Semantic Rules with Added Feature (Method 2) This approach uses the semantic role labels to extract a set of special translation rules, that on the target side form the smallest tree fragments in which one predicate and all of its core arguments are present. [sent-63, score-1.578]
37 These rules model the complete semantic structure of each predicate, and are used by the decoder in addition to the normal GHKM rules, which are extracted separately. [sent-64, score-0.631]
38 Starting by semantic role labeling the target parse trees, we modify the GHKM component of the system to extract a semantic rule for each predicate. [sent-65, score-0.998]
39 We define labels p as the set of semantic role labels related to predicate p. [sent-66, score-1.008]
40 That includes all 420 Nduevmber of rutelesst M Bae tsh eol idn e 12 14231946204137915 41 43 24069 05178509 Table 1: The number of the translation rules used by the three experimented methods of the labels of the arguments of p, and the label of p itself. [sent-67, score-0.791]
41 Then we add the following condition to the definition of the “frontier node” defined in Galley et al. [sent-68, score-0.043]
42 (2004): A frontier node must have either all or none of the semantic role labels from labels p in its descendants in the tree. [sent-69, score-0.829]
43 Adding this new condition, we extract one se- mantic rule for each predicate, and for that rule we discard the labels related to the other predicates. [sent-70, score-0.472]
44 This semantic rule will then have on its target side, the smallest tree fragment that contains all of the arguments of predicate p and the predicate itself. [sent-71, score-1.38]
45 Figure 2 depicts an example of a complete semantic rule. [sent-72, score-0.296]
46 Numbers following grammatical categories (for example, S-8 at the root) are the refined nonterminals produced by the split-merge parser. [sent-73, score-0.146]
47 In general, the tree side of the rule may extend below the nodes with semantic role labels because of the general constraint on frontier nodes that they must have a continuous span in the source (Chinese) side. [sent-74, score-1.096]
48 Also, the internal nodes of the rules (such as a node with PRED label in Figure 2) are removed because they are not used in decoding. [sent-75, score-0.352]
49 We also extract the regular GHKM rules using the original definition of the frontier nodes, and add the semantic rules to them. [sent-76, score-1.045]
50 To differentiate the semantic rules from the non-semantic ones, we add a new binary feature that is set to 1 for the semantic rules and to 0 for the rest of the rules. [sent-77, score-1.037]
51 3 Experiments Semantic role labeling was done using the PropBank standard (Palmer et al. [sent-78, score-0.292]
52 Our labeler uses a maximum entropy classifier and for identification and classification of semantic roles, and has a percision of 90% and a recall of 88%. [sent-80, score-0.3]
53 The features used for training the labeler are a subset of the features used by Gildea and Jurafsky (2000), Xue and Palmer (2004), and Pradhan et al. [sent-81, score-0.072]
54 The corpus was drawn from the newswire texts available from LDC. [sent-85, score-0.038]
55 1 We used a 392-sentence development set with four references for parameter tuning, and a 428-sentence test set with four references for testing. [sent-86, score-0.047]
56 They are drawn from the newswire portion of NIST evaluation (2004, 2005, 2006). [sent-87, score-0.038]
57 The development set and the test set only had sentences with less than 30 words for decoding speed. [sent-88, score-0.115]
58 , 2003), length penalty, and number of rules used, was used for the experiments. [sent-90, score-0.269]
59 In all of our experiments, we used the split-merge parsing method of Petrov et al. [sent-91, score-0.056]
60 on the training corpus, and mapped the semantic roles from the original trees to the result of the split-merge parser. [sent-92, score-0.488]
61 We used a syntax-based decoder with Earley parsing and cube pruning (Chiang, 2007). [sent-93, score-0.066]
62 We used the Minimum Error Rate Training (Och, 2003) to tune the decoding parameters for the development set and tested the best weights that were found on the test set. [sent-94, score-0.115]
63 We ran three sets of experiments: Baseline experiments, where we did not do any semantic role labeling prior to rule extraction and only extracted regular GHKM rules, experiments with our method of Section 2. [sent-95, score-0.828]
64 4 (Method 1), and a set of experiments with our method of Section 2. [sent-96, score-0.056]
65 Table 1 contains the numbers of the GHKM translation rules used by our three method. [sent-98, score-0.445]
66 The rules were filtered by the development and the test to increase the decoding speed. [sent-99, score-0.384]
67 The increases in the number of rules were expected, but they were not big enough to significantly change the performance of the decoder. [sent-100, score-0.269]
68 1 Results For every set of experiments, we ran MERT on the development set with 8 different starting weight vectors picked randomly. [sent-102, score-0.047]
69 We then tested the system on the test set, using for each experiment the weight vector from the iteration of MERT with the maximum BLEU score on the development set. [sent-104, score-0.047]
70 Table 3 shows the BLEU scores that we found on the test set, and their corresponding scores on the development set. [sent-105, score-0.047]
71 The language model is trained on the English side of entire data (1. [sent-107, score-0.062]
72 ) 421 RBSMeoaesufte hrlceio n dec2etc解o as决n roenosl v1t3levrte 亿hlyeth人oeprnio的 sob thlue 问emrosf题o,f1ca,1. [sent-110, score-0.041]
73 BMaestheloidne 2a r a b l e a g u e i s t h e b e s t pwairt hne dre tmo dcirsactuics rse tfhoerm mi idnd tlehe e amstid rdegleio enas dte rmegoicornat i nc t rhe fo drimsc wusisthio tnhe of u tnhiete udn sit aetdes st ,a hte s ,sa hide s . [sent-121, score-0.196]
74 Table 2: Comparison of example translations from the baseline method and our Method 2. [sent-123, score-0.148]
75 Method 1 system seems to behave slightly worse than the baseline and Method 2. [sent-126, score-0.053]
76 The reason for this behavior is that the rules that were extracted from the semantic role labeled corpus could have isolated semantic roles in them which would not necessarily get connected to the right predicate or argument during decoding. [sent-127, score-1.534]
77 In other words, it is possible for a rule to only contain one or some of the semantic arguments of a predicate, and not even include the predicate itself, and therefore there is no guarantee that the predicate will be translated with the right arguments and in the right order. [sent-128, score-1.461]
78 The difference between the BLEU scores of the best Method 2 results and the baseline is 0. [sent-129, score-0.053]
79 032) and it shows that incorporating semantic roles in machine translation is an effective approach. [sent-132, score-0.652]
80 Table 2 compares some translations from the baseline decoder and our Method 2. [sent-133, score-0.158]
81 The last two lines compare the baseline and Method 2. [sent-135, score-0.053]
82 These examples show how our Method 2 can outperform the baseline method, by translating complete semantic structures, and generating the semantic roles in the correct order in the target language. [sent-136, score-0.843]
83 In the first example, the predicate rely on for the argument themselves was not translated by the baseline decoder, but it was correctly translated by Method 2. [sent-137, score-0.514]
84 The second example is a case where the baseline method generated the arguments in the wrong order (in the case of facing and development), but the translation by Method 2 has the correct order. [sent-138, score-0.468]
85 In the last example we see that the arguments of the predicate discuss have the wrong order in the baseline translation, dBeLvEU Scteorset M Bae tshe lo ind e 21 2 6 . [sent-139, score-0.552]
86 980240 Table 3: BLEU scores on the test and development sets, of 8 experiments with random initial feature weights. [sent-142, score-0.047]
87 4 Conclusion We proposed two methods for incorporating semantic role labels in a string-to-tree machine translation system, by learning translation rules that are semantically enriched. [sent-144, score-1.279]
88 The first approach did not perform any better than the baseline, which we explained as being due to having rules with only partial semantic struc- tures and not having a way to guarantee that those rules will be used with each other in the right way. [sent-146, score-0.806]
89 The second approach significantly outperformed the baseline of our experiments, which shows that complete predicate-argument structures can improve the quality of machine translation. [sent-147, score-0.121]
90 The Proposition Bank: An annotated corpus of semantic roles. [sent-194, score-0.228]
91 BLEU: A method for automatic evaluation of machine translation. [sent-203, score-0.056]
wordName wordTfidf (topN-words)
[('predicate', 0.316), ('ghkm', 0.288), ('rules', 0.269), ('semantic', 0.228), ('role', 0.218), ('roles', 0.21), ('arguments', 0.183), ('translation', 0.176), ('rule', 0.155), ('gildea', 0.138), ('frontier', 0.137), ('labels', 0.123), ('galley', 0.115), ('nonterminals', 0.105), ('bleu', 0.099), ('bae', 0.094), ('npb', 0.094), ('palmer', 0.092), ('predicateargument', 0.089), ('tree', 0.087), ('daniel', 0.085), ('lending', 0.083), ('everybody', 0.077), ('labeling', 0.074), ('labelers', 0.072), ('opened', 0.072), ('labeler', 0.072), ('smt', 0.072), ('enriched', 0.069), ('door', 0.069), ('srikumar', 0.069), ('complete', 0.068), ('decoding', 0.068), ('decoder', 0.066), ('argument', 0.065), ('side', 0.062), ('chung', 0.061), ('regular', 0.06), ('pred', 0.059), ('pradhan', 0.059), ('urstenau', 0.058), ('method', 0.056), ('xiong', 0.056), ('scfg', 0.056), ('target', 0.056), ('martha', 0.056), ('wu', 0.055), ('levin', 0.055), ('augmenting', 0.055), ('nianwen', 0.053), ('alternations', 0.053), ('baseline', 0.053), ('petrov', 0.053), ('baker', 0.052), ('attach', 0.051), ('semantically', 0.051), ('core', 0.051), ('trees', 0.05), ('development', 0.047), ('rochester', 0.047), ('add', 0.043), ('nodes', 0.043), ('rhe', 0.041), ('eot', 0.041), ('biennial', 0.041), ('tagyoung', 0.041), ('hacioglu', 0.041), ('kadri', 0.041), ('nil', 0.041), ('warn', 0.041), ('licheng', 0.041), ('sob', 0.041), ('bloodgood', 0.041), ('derivable', 0.041), ('grafting', 0.041), ('hide', 0.041), ('marzieh', 0.041), ('oii', 0.041), ('refined', 0.041), ('xue', 0.041), ('mert', 0.041), ('adding', 0.041), ('guarantee', 0.04), ('translated', 0.04), ('label', 0.04), ('smallest', 0.039), ('extract', 0.039), ('translations', 0.039), ('newswire', 0.038), ('hte', 0.038), ('sot', 0.038), ('rse', 0.038), ('encompass', 0.038), ('hti', 0.038), ('hne', 0.038), ('vbg', 0.038), ('splitmerge', 0.038), ('incorporating', 0.038), ('extraction', 0.037), ('toutanova', 0.037)]
simIndex simValue paperId paperTitle
same-paper 1 0.99999976 314 acl-2013-Semantic Roles for String to Tree Machine Translation
Author: Marzieh Bazrafshan ; Daniel Gildea
Abstract: We experiment with adding semantic role information to a string-to-tree machine translation system based on the rule extraction procedure of Galley et al. (2004). We compare methods based on augmenting the set of nonterminals by adding semantic role labels, and altering the rule extraction process to produce a separate set of rules for each predicate that encompass its entire predicate-argument structure. Our results demonstrate that the second approach is effective in increasing the quality of translations.
2 0.22859861 306 acl-2013-SPred: Large-scale Harvesting of Semantic Predicates
Author: Tiziano Flati ; Roberto Navigli
Abstract: We present SPred, a novel method for the creation of large repositories of semantic predicates. We start from existing collocations to form lexical predicates (e.g., break ∗) and learn the semantic classes that best f∗it) tahned ∗ argument. Taon idco this, we extract failtl thhee ∗ occurrences ion Wikipedia ewxthraiccht match the predicate and abstract its arguments to general semantic classes (e.g., break BODY PART, break AGREEMENT, etc.). Our experiments show that we are able to create a large collection of semantic predicates from the Oxford Advanced Learner’s Dictionary with high precision and recall, and perform well against the most similar approach.
3 0.20509876 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
4 0.20187643 27 acl-2013-A Two Level Model for Context Sensitive Inference Rules
Author: Oren Melamud ; Jonathan Berant ; Ido Dagan ; Jacob Goldberger ; Idan Szpektor
Abstract: Automatic acquisition of inference rules for predicates has been commonly addressed by computing distributional similarity between vectors of argument words, operating at the word space level. A recent line of work, which addresses context sensitivity of rules, represented contexts in a latent topic space and computed similarity over topic vectors. We propose a novel two-level model, which computes similarities between word-level vectors that are biased by topic-level context representations. Evaluations on a naturallydistributed dataset show that our model significantly outperforms prior word-level and topic-level models. We also release a first context-sensitive inference rule set.
5 0.19927721 361 acl-2013-Travatar: A Forest-to-String Machine Translation Engine based on Tree Transducers
Author: Graham Neubig
Abstract: In this paper we describe Travatar, a forest-to-string machine translation (MT) engine based on tree transducers. It provides an open-source C++ implementation for the entire forest-to-string MT pipeline, including rule extraction, tuning, decoding, and evaluation. There are a number of options for model training, and tuning includes advanced options such as hypergraph MERT, and training of sparse features through online learning. The training pipeline is modeled after that of the popular Moses decoder, so users familiar with Moses should be able to get started quickly. We perform a validation experiment of the decoder on EnglishJapanese machine translation, and find that it is possible to achieve greater accuracy than translation using phrase-based and hierarchical-phrase-based translation. As auxiliary results, we also compare different syntactic parsers and alignment techniques that we tested in the process of developing the decoder. Travatar is available under the LGPL at http : / /phont ron . com/t ravat ar
6 0.18929875 189 acl-2013-ImpAr: A Deterministic Algorithm for Implicit Semantic Role Labelling
7 0.17981742 320 acl-2013-Shallow Local Multi-Bottom-up Tree Transducers in Statistical Machine Translation
8 0.17131391 195 acl-2013-Improving machine translation by training against an automatic semantic frame based evaluation metric
9 0.15083544 226 acl-2013-Learning to Prune: Context-Sensitive Pruning for Syntactic MT
10 0.14886637 19 acl-2013-A Shift-Reduce Parsing Algorithm for Phrase-based String-to-Dependency Translation
11 0.13947988 166 acl-2013-Generalized Reordering Rules for Improved SMT
12 0.13827664 200 acl-2013-Integrating Phrase-based Reordering Features into a Chart-based Decoder for Machine Translation
13 0.13523395 285 acl-2013-Propminer: A Workflow for Interactive Information Extraction and Exploration using Dependency Trees
14 0.13376676 376 acl-2013-Using Lexical Expansion to Learn Inference Rules from Sparse Data
15 0.13267528 267 acl-2013-PARMA: A Predicate Argument Aligner
16 0.12815107 312 acl-2013-Semantic Parsing as Machine Translation
17 0.12624267 11 acl-2013-A Multi-Domain Translation Model Framework for Statistical Machine Translation
18 0.12346756 46 acl-2013-An Infinite Hierarchical Bayesian Model of Phrasal Translation
19 0.12262894 223 acl-2013-Learning a Phrase-based Translation Model from Monolingual Data with Application to Domain Adaptation
20 0.12253701 10 acl-2013-A Markov Model of Machine Translation using Non-parametric Bayesian Inference
topicId topicWeight
[(0, 0.297), (1, -0.161), (2, 0.078), (3, -0.005), (4, -0.133), (5, 0.14), (6, -0.028), (7, 0.073), (8, 0.029), (9, 0.114), (10, 0.031), (11, 0.142), (12, 0.099), (13, 0.097), (14, 0.065), (15, -0.047), (16, 0.165), (17, 0.044), (18, 0.184), (19, 0.111), (20, 0.003), (21, 0.092), (22, -0.089), (23, 0.032), (24, 0.034), (25, -0.016), (26, 0.114), (27, -0.054), (28, 0.052), (29, 0.026), (30, 0.007), (31, -0.08), (32, 0.04), (33, 0.01), (34, 0.021), (35, -0.018), (36, -0.052), (37, -0.048), (38, 0.034), (39, -0.003), (40, -0.044), (41, 0.076), (42, 0.033), (43, 0.018), (44, 0.065), (45, 0.078), (46, -0.021), (47, -0.023), (48, 0.036), (49, -0.023)]
simIndex simValue paperId paperTitle
same-paper 1 0.96009123 314 acl-2013-Semantic Roles for String to Tree Machine Translation
Author: Marzieh Bazrafshan ; Daniel Gildea
Abstract: We experiment with adding semantic role information to a string-to-tree machine translation system based on the rule extraction procedure of Galley et al. (2004). We compare methods based on augmenting the set of nonterminals by adding semantic role labels, and altering the rule extraction process to produce a separate set of rules for each predicate that encompass its entire predicate-argument structure. Our results demonstrate that the second approach is effective in increasing the quality of translations.
2 0.77290267 320 acl-2013-Shallow Local Multi-Bottom-up Tree Transducers in Statistical Machine Translation
Author: Fabienne Braune ; Nina Seemann ; Daniel Quernheim ; Andreas Maletti
Abstract: We present a new translation model integrating the shallow local multi bottomup tree transducer. We perform a largescale empirical evaluation of our obtained system, which demonstrates that we significantly beat a realistic tree-to-tree baseline on the WMT 2009 English → German tlriannes olnati tohne tWasMk.T TA 2s0 an a Edndgitliisonha →l c Gonetrrmibauntion we make the developed software and complete tool-chain publicly available for further experimentation.
3 0.72064286 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 0.70058364 361 acl-2013-Travatar: A Forest-to-String Machine Translation Engine based on Tree Transducers
Author: Graham Neubig
Abstract: In this paper we describe Travatar, a forest-to-string machine translation (MT) engine based on tree transducers. It provides an open-source C++ implementation for the entire forest-to-string MT pipeline, including rule extraction, tuning, decoding, and evaluation. There are a number of options for model training, and tuning includes advanced options such as hypergraph MERT, and training of sparse features through online learning. The training pipeline is modeled after that of the popular Moses decoder, so users familiar with Moses should be able to get started quickly. We perform a validation experiment of the decoder on EnglishJapanese machine translation, and find that it is possible to achieve greater accuracy than translation using phrase-based and hierarchical-phrase-based translation. As auxiliary results, we also compare different syntactic parsers and alignment techniques that we tested in the process of developing the decoder. Travatar is available under the LGPL at http : / /phont ron . com/t ravat ar
5 0.68883801 189 acl-2013-ImpAr: A Deterministic Algorithm for Implicit Semantic Role Labelling
Author: Egoitz Laparra ; German Rigau
Abstract: This paper presents a novel deterministic algorithm for implicit Semantic Role Labeling. The system exploits a very simple but relevant discursive property, the argument coherence over different instances of a predicate. The algorithm solves the implicit arguments sequentially, exploiting not only explicit but also the implicit arguments previously solved. In addition, we empirically demonstrate that the algorithm obtains very competitive and robust performances with respect to supervised approaches that require large amounts of costly training data.
6 0.66760772 180 acl-2013-Handling Ambiguities of Bilingual Predicate-Argument Structures for Statistical Machine Translation
7 0.63222247 285 acl-2013-Propminer: A Workflow for Interactive Information Extraction and Exploration using Dependency Trees
8 0.62781167 376 acl-2013-Using Lexical Expansion to Learn Inference Rules from Sparse Data
9 0.62357605 27 acl-2013-A Two Level Model for Context Sensitive Inference Rules
10 0.61969435 195 acl-2013-Improving machine translation by training against an automatic semantic frame based evaluation metric
11 0.61555594 306 acl-2013-SPred: Large-scale Harvesting of Semantic Predicates
12 0.61181521 226 acl-2013-Learning to Prune: Context-Sensitive Pruning for Syntactic MT
13 0.60308456 267 acl-2013-PARMA: A Predicate Argument Aligner
14 0.59801918 200 acl-2013-Integrating Phrase-based Reordering Features into a Chart-based Decoder for Machine Translation
15 0.59796095 165 acl-2013-General binarization for parsing and translation
16 0.58502966 16 acl-2013-A Novel Translation Framework Based on Rhetorical Structure Theory
17 0.58447778 330 acl-2013-Stem Translation with Affix-Based Rule Selection for Agglutinative Languages
18 0.57395786 17 acl-2013-A Random Walk Approach to Selectional Preferences Based on Preference Ranking and Propagation
19 0.5586068 137 acl-2013-Enlisting the Ghost: Modeling Empty Categories for Machine Translation
20 0.5566408 110 acl-2013-Deepfix: Statistical Post-editing of Statistical Machine Translation Using Deep Syntactic Analysis
topicId topicWeight
[(0, 0.041), (6, 0.018), (11, 0.062), (14, 0.266), (24, 0.029), (26, 0.054), (35, 0.067), (42, 0.089), (48, 0.073), (70, 0.046), (88, 0.017), (90, 0.079), (95, 0.09)]
simIndex simValue paperId paperTitle
1 0.85713053 357 acl-2013-Transfer Learning for Constituency-Based Grammars
Author: Yuan Zhang ; Regina Barzilay ; Amir Globerson
Abstract: In this paper, we consider the problem of cross-formalism transfer in parsing. We are interested in parsing constituencybased grammars such as HPSG and CCG using a small amount of data specific for the target formalism, and a large quantity of coarse CFG annotations from the Penn Treebank. While all of the target formalisms share a similar basic syntactic structure with Penn Treebank CFG, they also encode additional constraints and semantic features. To handle this apparent discrepancy, we design a probabilistic model that jointly generates CFG and target formalism parses. The model includes features of both parses, allowing trans- fer between the formalisms, while preserving parsing efficiency. We evaluate our approach on three constituency-based grammars CCG, HPSG, and LFG, augmented with the Penn Treebank-1. Our experiments show that across all three formalisms, the target parsers significantly benefit from the coarse annotations.1 —
same-paper 2 0.80046058 314 acl-2013-Semantic Roles for String to Tree Machine Translation
Author: Marzieh Bazrafshan ; Daniel Gildea
Abstract: We experiment with adding semantic role information to a string-to-tree machine translation system based on the rule extraction procedure of Galley et al. (2004). We compare methods based on augmenting the set of nonterminals by adding semantic role labels, and altering the rule extraction process to produce a separate set of rules for each predicate that encompass its entire predicate-argument structure. Our results demonstrate that the second approach is effective in increasing the quality of translations.
3 0.75485712 13 acl-2013-A New Syntactic Metric for Evaluation of Machine Translation
Author: Melania Duma ; Cristina Vertan ; Wolfgang Menzel
Abstract: Machine translation (MT) evaluation aims at measuring the quality of a candidate translation by comparing it with a reference translation. This comparison can be performed on multiple levels: lexical, syntactic or semantic. In this paper, we propose a new syntactic metric for MT evaluation based on the comparison of the dependency structures of the reference and the candidate translations. The dependency structures are obtained by means of a Weighted Constraints Dependency Grammar parser. Based on experiments performed on English to German translations, we show that the new metric correlates well with human judgments at the system level. 1
4 0.75069124 31 acl-2013-A corpus-based evaluation method for Distributional Semantic Models
Author: Abdellah Fourtassi ; Emmanuel Dupoux
Abstract: Evaluation methods for Distributional Semantic Models typically rely on behaviorally derived gold standards. These methods are difficult to deploy in languages with scarce linguistic/behavioral resources. We introduce a corpus-based measure that evaluates the stability of the lexical semantic similarity space using a pseudo-synonym same-different detection task and no external resources. We show that it enables to predict two behaviorbased measures across a range of parameters in a Latent Semantic Analysis model.
5 0.7314018 137 acl-2013-Enlisting the Ghost: Modeling Empty Categories for Machine Translation
Author: Bing Xiang ; Xiaoqiang Luo ; Bowen Zhou
Abstract: Empty categories (EC) are artificial elements in Penn Treebanks motivated by the government-binding (GB) theory to explain certain language phenomena such as pro-drop. ECs are ubiquitous in languages like Chinese, but they are tacitly ignored in most machine translation (MT) work because of their elusive nature. In this paper we present a comprehensive treatment of ECs by first recovering them with a structured MaxEnt model with a rich set of syntactic and lexical features, and then incorporating the predicted ECs into a Chinese-to-English machine translation task through multiple approaches, including the extraction of EC-specific sparse features. We show that the recovered empty categories not only improve the word alignment quality, but also lead to significant improvements in a large-scale state-of-the-art syntactic MT system.
6 0.61415732 4 acl-2013-A Context Free TAG Variant
7 0.6089499 80 acl-2013-Chinese Parsing Exploiting Characters
8 0.60379809 155 acl-2013-Fast and Accurate Shift-Reduce Constituent Parsing
9 0.59856921 226 acl-2013-Learning to Prune: Context-Sensitive Pruning for Syntactic MT
10 0.59438932 275 acl-2013-Parsing with Compositional Vector Grammars
11 0.5939604 303 acl-2013-Robust multilingual statistical morphological generation models
12 0.58812183 267 acl-2013-PARMA: A Predicate Argument Aligner
13 0.5878846 335 acl-2013-Survey on parsing three dependency representations for English
14 0.58466911 38 acl-2013-Additive Neural Networks for Statistical Machine Translation
15 0.58458626 174 acl-2013-Graph Propagation for Paraphrasing Out-of-Vocabulary Words in Statistical Machine Translation
16 0.58429199 205 acl-2013-Joint Apposition Extraction with Syntactic and Semantic Constraints
17 0.58326519 264 acl-2013-Online Relative Margin Maximization for Statistical Machine Translation
18 0.58193272 144 acl-2013-Explicit and Implicit Syntactic Features for Text Classification
19 0.58164251 57 acl-2013-Arguments and Modifiers from the Learner's Perspective
20 0.58134669 223 acl-2013-Learning a Phrase-based Translation Model from Monolingual Data with Application to Domain Adaptation