{ "paper_id": "Y13-1026", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T13:31:58.890142Z" }, "title": "Effects of Parsing Errors on Pre-reordering Performance for Chinese-to-Japanese SMT", "authors": [ { "first": "Dan", "middle": [], "last": "Han", "suffix": "", "affiliation": { "laboratory": "", "institution": "The Graduate University For Advanced Studies", "location": {} }, "email": "handan@nii.ac.jp" }, { "first": "Pascual", "middle": [], "last": "Mart\u00ednez-G\u00f3mez", "suffix": "", "affiliation": { "laboratory": "", "institution": "National Institute of Informatics", "location": {} }, "email": "pascual@nii.ac.jp" }, { "first": "Yusuke", "middle": [], "last": "Miyao", "suffix": "", "affiliation": { "laboratory": "", "institution": "The Graduate University For Advanced Studies", "location": {} }, "email": "yusuke@nii.ac.jp" }, { "first": "Katsuhito", "middle": [], "last": "Sudoh", "suffix": "", "affiliation": { "laboratory": "", "institution": "NTT Corporation", "location": {} }, "email": "sudoh.katsuhito@lab.ntt.co.jp" }, { "first": "Masaaki", "middle": [], "last": "Nagata", "suffix": "", "affiliation": { "laboratory": "", "institution": "NTT Corporation", "location": {} }, "email": "nagata.masaaki@lab.ntt.co.jp" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "Linguistically motivated reordering methods have been developed to improve word alignment especially for Statistical Machine Translation (SMT) on long distance language pairs. However, since they highly rely on the parsing accuracy, it is useful to explore the relationship between parsing and reordering. For Chinese-to-Japanese SMT, we carry out a three-stage incremental comparative analysis to observe the effects of different parsing errors on reordering performance by combining empirical and descriptive approaches. For the empirical approach, we quantify the distribution of general parsing errors along with reordering qualities whereas for the descriptive approach, we extract seven influential error patterns and examine their correlation with reordering errors.", "pdf_parse": { "paper_id": "Y13-1026", "_pdf_hash": "", "abstract": [ { "text": "Linguistically motivated reordering methods have been developed to improve word alignment especially for Statistical Machine Translation (SMT) on long distance language pairs. However, since they highly rely on the parsing accuracy, it is useful to explore the relationship between parsing and reordering. For Chinese-to-Japanese SMT, we carry out a three-stage incremental comparative analysis to observe the effects of different parsing errors on reordering performance by combining empirical and descriptive approaches. For the empirical approach, we quantify the distribution of general parsing errors along with reordering qualities whereas for the descriptive approach, we extract seven influential error patterns and examine their correlation with reordering errors.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "Statistical machine translation is a challenging and well established task in the community of computational linguistics. One of the key components of statistical machine translation systems are word alignment techniques, where the words from sentences in a source language are mapped to words from sentences in a target language. When estimating the most appropriate word alignments, it is unfeasible to explore every possible word correspondence due to the combinatorial complexity. Considering local permutations of words might be effective to translate languages with a similar sentence structure, but these methods have a limited performance when translating sentences from languages with different syntactical structures.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "An effective technique to translate sentences between distant language pairs is pre-reordering, where words in sentences from the source language are re-arranged with the objective to resemble the word order of the target language. Rearranging rules are automatically extracted (Xia and McCord, 2004; Genzel, 2010) , or linguistically motivated (Xu et al., 2009; Isozaki et al., 2010; Han et al., 2012; Han et al., 2013) . We work following the latter strategy, where the source sentence is parsed to find its syntactical structure, and linguistically-motivated rules are used in combination with the structure of the sentence to guide the word reordering. The language pair under consideration is Chinese-to-Japanese, which despite their common roots, it is a well known language pair for their different sentence structure.", "cite_spans": [ { "start": 278, "end": 300, "text": "(Xia and McCord, 2004;", "ref_id": "BIBREF17" }, { "start": 301, "end": 314, "text": "Genzel, 2010)", "ref_id": "BIBREF2" }, { "start": 345, "end": 362, "text": "(Xu et al., 2009;", "ref_id": "BIBREF18" }, { "start": 363, "end": 384, "text": "Isozaki et al., 2010;", "ref_id": "BIBREF9" }, { "start": 385, "end": 402, "text": "Han et al., 2012;", "ref_id": "BIBREF5" }, { "start": 403, "end": 420, "text": "Han et al., 2013)", "ref_id": "BIBREF6" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "However, syntax-based pre-reordering techniques are sensitive to parsing errors, but insight into their relationship has been elusive. The contribution of this work is two fold. First, we provide an empirical analysis where we quantify the aggregated impact of parsing errors on pre-reordering performance. Second, we define seven patterns of the most common and influential parsing errors and we carry out a descriptive analysis to examine their relationship with reordering errors. We combine an empirical and descriptive approach to present a three-stage incremental comparative analysis to observe the effect of different parsing errors on reordering performance.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In Section 2, after a brief description on the prereordering method that we use for experiments, we will introduce some related works on parsing error analysis and analysis on the relation between parsing and machine translation. From a general perspective, we describe our analysis methods for this work in Section 3. Then, we carry out the analysis and exhibit the results in Section 4 and Section 5. The last two sections are dedicated to discussion, future directions and summarize our findings.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Vb-H VV VE VC VA P BEI LB SB RM-D NN NR NT PN OD CD M FW CC ETC LC DEV DT JJ SP IJ ON Table 1 : Lists of POS tags for identifying words as Vb-H, RM-D, and BEI. (Han et al., 2013) 2 Background", "cite_spans": [ { "start": 160, "end": 178, "text": "(Han et al., 2013)", "ref_id": "BIBREF6" } ], "ref_spans": [ { "start": 86, "end": 93, "text": "Table 1", "ref_id": null } ], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Since local reordering models which are integrated in phrase-based SMT systems do not perform well for distant language pairs due to their different syntactic structures, pre-reordering methods have been proposed to supply the need for improving the word alignment. Han et al. (2013) described one of the latest pre-reordering methods (DPC) which was based on dependency parsing. The authors were using an unlabeled dependency parser to extract the syntactic information of Chinese sentences, and then by combining with part-of-speech (POS) tags 1 , they defined a set of heuristic reordering rules to guide the reordering. The essential idea of DPC is to move so-called verbal block (Vb) 2 to the right-hand side of its right-most dependent (RM-D) for a Subject-Verb-Object (SVO) language to resemble a Subject-Object-Verb (SOV) language's word order. Table 1 shows the POS tags that are used to identify words as Vb-H, RM-D, or BEI (a Vb-H involves in a bei-construction) in a sentence from Han et al. (2013) . Figure 1 shows an example of unlabeled dependency parse tree of a Chinese sentence aligned with its Japanese translation. According to the reordering method, \"went\" will be reordered behind of \"bookstore\" while \"buy -ed\" will be reordered to the right-hand side of \"book\", and thus the sentence will follow a SOV word order as Japanese. However, if \"book\" was wrongly recognized as the dependent of \"went\" in the dependency structure, \"went\" will be wrongly reordered to the righthand side of \"book\". Therefore, syntactic structure based reordering methods highly rely on the parsing accuracy. In order to further improve word alignments or refine existing reordering models, it", "cite_spans": [ { "start": 266, "end": 283, "text": "Han et al. (2013)", "ref_id": "BIBREF6" }, { "start": 993, "end": 1010, "text": "Han et al. (2013)", "ref_id": "BIBREF6" } ], "ref_spans": [ { "start": 853, "end": 860, "text": "Table 1", "ref_id": null }, { "start": 1013, "end": 1021, "text": "Figure 1", "ref_id": null } ], "eq_spans": [], "section": "Reordering Model", "sec_num": "2.1" }, { "text": ". . . . PinYin: . . ta1 . . qu4 . . shu1dian4 . . mai3 . . le5 . . yi1 . . ben3 . . shu1 . . . . Chinese: . . \u4ed6 . . \u53bb . . \u4e66\u5e97 . . \u4e70 . . \u4e86 . . \u4e00 . . \u672c . . \u4e66 . . \u3002 . . English: . .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Reordering Model", "sec_num": "2.1" }, { "text": "He . . went (to) . . bookstore . .", "cite_spans": [ { "start": 12, "end": 16, "text": "(to)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Reordering Model", "sec_num": "2.1" }, { "text": ". . Figure 1 : Example of unlabeled dependency parse tree of a Chinese sentence (SVO) with word aligned to its Japanese counterpart (SOV). Arrows are pointing from heads to dependents.", "cite_spans": [], "ref_spans": [ { "start": 4, "end": 12, "text": "Figure 1", "ref_id": null } ], "eq_spans": [], "section": "buy", "sec_num": null }, { "text": "-ed . . a . . . . book . . . . ROOT . . o . . o . . o . . o . . o . . o . . o . . o . Japanese: . \u5f7c (\u306f) . \u672c\u5c4b (\u306b) . \u884c\u3063\u3066 . \u672c (\u3092) . \u8cb7\u3063 . \u305f . \u3002", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "buy", "sec_num": null }, { "text": "is important to observe the effects of parsing errors on reordering performance.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "buy", "sec_num": null }, { "text": "In this analysis, we borrow this state-of-the-art pre-reordering model for our experiments since it is a rule-based pre-reordering method for a distant language pair based on dependency parsing as well as its extensibility to other language pairs.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "buy", "sec_num": null }, { "text": "Although there are studies on analyzing parsing errors and reordering errors, as far as we know, there is not any work on observing the relationship between these two types of errors.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2.2" }, { "text": "One most relevant work to ours is observing the impact of parsing accuracy on a SMT system introduced in Quirk and Corston-Oliver (2006) . They showed the general idea that syntax-based SMT models are sensitive to syntactic analysis. However, they did not further analyze concrete parsing error types that affect task accuracy. Green (2011) explored the effects of noun phrase bracketing in dependency parsing in English, and further on English to Czech machine translation. But the work focused on using noun phrase structure to improve a machine translation framework. In the work of Katz-Brown et al. (2011), they proposed a training method to improve a parser's performance by using reordering quality to examine the parse quality. But they did not study the relationship between reordering quality and parse quality.", "cite_spans": [ { "start": 105, "end": 136, "text": "Quirk and Corston-Oliver (2006)", "ref_id": "BIBREF15" }, { "start": 328, "end": 340, "text": "Green (2011)", "ref_id": "BIBREF4" } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2.2" }, { "text": "There are more works on parsing error analysis. For instance, Hara et al. (2009) defined several types of parsing error patterns on predicate argument relation and tested them with a Headdriven phrase structure grammar (HPSG) (Pollard and Sag, 1994) parser (Miyao and Tsujii, 2008) . McDonald and Nivre (2007) explored parsing errors for data-driven dependency parsing by comparing a graph-based parser with a transitionbased parser, which are representing two dominant parsing models. At the same time, Dredze et al. (2007) provided a comparison analysis on differences in annotation guidelines among treebanks which were suspected to be responsible for dependency parsing errors in domain adaptation tasks. Unlike analyzing parsing errors, authors in Yu et al. (2011) focused on the difficulties in Chinese deep parsing by comparing the linguistic properties between Chinese and English.", "cite_spans": [ { "start": 62, "end": 80, "text": "Hara et al. (2009)", "ref_id": "BIBREF7" }, { "start": 226, "end": 249, "text": "(Pollard and Sag, 1994)", "ref_id": "BIBREF14" }, { "start": 257, "end": 281, "text": "(Miyao and Tsujii, 2008)", "ref_id": "BIBREF13" }, { "start": 284, "end": 309, "text": "McDonald and Nivre (2007)", "ref_id": "BIBREF12" }, { "start": 504, "end": 524, "text": "Dredze et al. (2007)", "ref_id": "BIBREF0" }, { "start": 753, "end": 769, "text": "Yu et al. (2011)", "ref_id": "BIBREF19" } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2.2" }, { "text": "There are also works on reordering error analysis like Han et al. (2012) which examined an existing reordering method and refined it after a detailed linguistic analysis on reordering issues. Although they discovered that parsing errors affect the reordering quality, they did not observe the concrete relationship. On the other hand, Gim\u00e9nez and M\u00e0rquez (2008) proposed an automatic error analysis method of machine translation output, by compiling a set of metric variants. However, they did not provide insight on what SMT component caused low translation performance.", "cite_spans": [ { "start": 55, "end": 72, "text": "Han et al. (2012)", "ref_id": "BIBREF5" }, { "start": 335, "end": 361, "text": "Gim\u00e9nez and M\u00e0rquez (2008)", "ref_id": "BIBREF3" } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2.2" }, { "text": "We combine an empirical approach with a descriptive approach to observe the effects of parsing errors on pre-reordering performance in three stages: preliminary experiment stage, POS tag level stage, and dependency type level stage. First, we provide a general idea of the sensitiveness of parsing errors on reordering method. Then, we use POS tags to identify parsing errors and quantify the aggregate impact on reordering performance. Finally, we define several concrete error patterns and examine their effects on reordering qualities.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis Method", "sec_num": "3" }, { "text": "In order to test for an upper bound of the reordering performance and examine the specific parsing errors that affect reordering, one way is to contrast the reordering based on error-free parse trees with the reordering based on auto-parse trees. Error-free parse trees are considered as gold trees.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis Method", "sec_num": "3" }, { "text": "In the preliminary experiment stage, we set up two benchmarks in two scenarios. For scenario 1, the benchmark is manually reordered Chinese sentence on the basis of Japanese reference. By measuring the word order similarities between the benchmark and the gold-tree based reordered sentence as well as between the benchmark and the auto-parse tree based reordered sentence separately, we quantify the extent of parsing errors that influence reordering. Meanwhile, the former measurement shows additionally the general figure of the upper bound of the reordering method. However, since it is not only time-consuming but also labor-intensive to set up the benchmark in scenario 1, we use the Japanese reference as the benchmark in scenario 2 and follow the same strategies as in scenario 1 to calculate the word order similarities. More detailed description on the preliminary experiment is given in Section 4.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis Method", "sec_num": "3" }, { "text": "In POS tag level stage, we compare the goldtree with auto-parse tree along with reordering quality to explore the relationship between general parsing errors and reordering from two aspects: the percentages of top three most frequent dependent's POS tags that point to wrong heads and the percentages of top two most frequent head's POS tags that are recognized wrongly. The percentages of other POS tags are not provided because they are negligible. Our objective is to profile general parsing errors' distribution. However, this does not imply that those errors are the cause of the reordering errors. Section 5.1 includes more concrete analysis results.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis Method", "sec_num": "3" }, { "text": "In dependency type level stage, we classify the most influential parsing errors on reordering into three superclasses and seven subclasses according to the methodology of the reordering method. We then plot the distribution of these parsing errors for various reordering qualities. In Section 5.2, we illustrate these parsing errors with examples.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis Method", "sec_num": "3" }, { "text": "In order to build up gold parse tree sets for comparison, we used the annotated sentences from Chinese Penn Treebank ver. 7.0 (CTB-7) which is a well known corpus that consists of parsed text in five genres. They are Chinese newswire (NS), magazine news (NM), broadcast news (BN), broadcast conversation programs (BC), and web newsgroups, weblogs (NW).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Gold Data", "sec_num": "4.1" }, { "text": "We first randomly selected 517 unique sentences (hereinafter set-1) from all five genres in development set of CTB-7 which is split according to . However, we found that sentences in BC and NW are mainly from spoken language, which tend to have faults like repetitions, incomplete sentences, corrections, or incorrect sentence segmentation. Therefore, we randomly selected another 2, 126 unique sentences (hereinafter set-2) within a limit to three genres: NS, NM, and BN. Table 2 shows the statistics of all selected sentences in five genres respectively. For converting CTB-7 parsed text to dependency parse trees, we used an open utility Penn2Malt 3 which converts Penn Treebank into MaltTab format containing dependency information. Since the head rules that Penn2Malt recommended for converting on its website do not contain three new annotation types in CTB-7, we added three new ones for them as follows: FLR (Fillers) and DFL (Disfluency) head on right-hand branch; INC (Incomplete sentences) follows the same head rule as FRAG (Fragment).", "cite_spans": [], "ref_spans": [ { "start": 473, "end": 480, "text": "Table 2", "ref_id": "TABREF1" } ], "eq_spans": [], "section": "Gold Data", "sec_num": "4.1" }, { "text": "Meanwhile, professional human translators translated all Chinese sentences in both set-1 and set-2 into Japanese. Thereafter, according to the Japanese references, Chinese sentences in set-1 have been manually reordered as the same word orders as their Japanese counterparts by a bilingual speaker of Chinese and Japanese for the experiments in scenario 1. For example, the Chinese sentence in Figure 1 is following the word order of \"He bookstore went (to) a book buy (-ed) .\" in the handcrafted reordered set since it resembles the Japanese word order.", "cite_spans": [], "ref_spans": [ { "start": 394, "end": 402, "text": "Figure 1", "ref_id": null } ], "eq_spans": [], "section": "Gold Data", "sec_num": "4.1" }, { "text": "We use Kendall's tau (\u03c4 ) rank correlation coefficient (Isozaki et al., 2010) to measure word order similarities between sentences in two different scenarios. In the first scenario, we use the set of manually reordered Chinese sentences from set-1 as benchmark and compare it with the set of automatically reordered Chinese sentences. In the second scenario, we combine set-1 and set-2 to obtain a larger data set. (Gao and Vogel, 2008) .", "cite_spans": [ { "start": 55, "end": 77, "text": "(Isozaki et al., 2010)", "ref_id": "BIBREF9" }, { "start": 415, "end": 436, "text": "(Gao and Vogel, 2008)", "ref_id": "BIBREF1" } ], "ref_spans": [], "eq_spans": [], "section": "Evaluation", "sec_num": "4.2" }, { "text": "In both scenarios, we carry out the reordering method DPC (See Section 2.1). Auto-parse trees are generated by an unlabeled Chinese dependency parser, Corbit 4 (Hatori et al., 2011) . Gold trees 5 are converted from CTB-7 parsed text which are created by human annotators. More specifically, we refer to auto-parse tree based reordering system as Auto-DPC and to gold-tree based reordering system as Gold-DPC. Baseline system uses unreordered Chinese sentences.", "cite_spans": [ { "start": 160, "end": 181, "text": "(Hatori et al., 2011)", "ref_id": "BIBREF8" } ], "ref_spans": [], "eq_spans": [], "section": "Evaluation", "sec_num": "4.2" }, { "text": "Scenario 1 Preliminary observation about the effects of parsing errors on reordering performance is to compare word order similarities between manually reordered Chinese sentences and automatically reordered Chinese sentences from set-1. Table 3 shows the average \u03c4 value.", "cite_spans": [], "ref_spans": [ { "start": 238, "end": 245, "text": "Table 3", "ref_id": "TABREF3" } ], "eq_spans": [], "section": "Evaluation", "sec_num": "4.2" }, { "text": "For baseline system, the average \u03c4 value shows how similar these 517 Chinese sentences between manually reordered ones and non-reordered ones are. Comparing with manually reordered Chinese, both Auto-DPC and Gold-DPC achieved higher average \u03c4 value than baseline, which imply that the reordering method DPC positively reordered the Chinese sentences and improved the word alignment. Nevertheless, a slightly lower average \u03c4 value of Auto-DPC shows that DPC is sensitive on parsing errors. This assumption is also confirmed by the average \u03c4 value between Auto-DPC and Gold-DPC. However, the difference of \u03c4 values are limited. We hence increase the test data by adding set-2 for further experiments in scenario 2.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Evaluation", "sec_num": "4.2" }, { "text": "Scenario 2 Since we do not have manually reordered Chinese sentences as benchmark for set-2, we calculate the Kendall's tau between Chinese sentences and their Japanese counterparts for both data sets by using the MGIZA++ alignment file, ch-ja.A3.final. The comparison implies how monotonically the Chinese sentences have been reordered to align with Japanese. We use MeCab 6 (Kudo and Matsumoto, 2000) to segment Japanese sentences and also filter out sentences with more than 64 tokens. There are 2, 236 valid Chinese-Japanese bilingual sentences in total. Figure 2 shows the distribution of Kendall's tau from three systems in which the baseline is built up by using ordinary Chinese. In Figure 2 , baseline system contains a large numbers of non-monotonic aligned sentences, whereas both Auto-DPC and Gold-DPC increased the amount of sentences that achieved high \u03c4 values. Reordering based on gold-tree reduced more percentage of low \u03c4 sentences than reordering based on automatically parsed trees. Especially, the amount of sentence difference in 0.9 < \u03c4 <= 1 between Gold-DPC and Auto-DPC shows that reordering method DPC has a high sensitivity on parsing errors, which enhances the conclusions from the preliminary observation in scenario 1. Furthermore, the performance of reordering system Gold-DPC sketches the figure of upper bound of the reordering method.", "cite_spans": [ { "start": 376, "end": 402, "text": "(Kudo and Matsumoto, 2000)", "ref_id": "BIBREF11" } ], "ref_spans": [ { "start": 559, "end": 567, "text": "Figure 2", "ref_id": "FIGREF0" }, { "start": 691, "end": 699, "text": "Figure 2", "ref_id": "FIGREF0" } ], "eq_spans": [], "section": "Evaluation", "sec_num": "4.2" }, { "text": "Preliminary experiments in Section 4 provide a general idea of the effects of parsing errors on reordering. In order to achieve more explicit relationship between specific parsing errors and reordering issues, we first identify concrete parsing errors by comparing gold-trees with auto-parse ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis on Causes of Reordering Errors", "sec_num": "5" }, { "text": ". ROOT . . o . . o . . o . . o . . o . . o . . o .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Analysis on Causes of Reordering Errors", "sec_num": "5" }, { "text": ". o Figure 3 : A possible wrong dependency parse tree of the example in Figure 1. trees. Since the syntactic information that guides reordering in DPC is limited to dependency structure and POS tags, for analysis on the causes of reordering errors, we examine parsing errors from these two linguistic categories. In this section, the value of Kendall's tau measures the word order similarity between Gold-DPC and Auto-DPC.", "cite_spans": [], "ref_spans": [ { "start": 4, "end": 12, "text": "Figure 3", "ref_id": null }, { "start": 72, "end": 81, "text": "Figure 1.", "ref_id": null } ], "eq_spans": [], "section": "Analysis on Causes of Reordering Errors", "sec_num": "5" }, { "text": "There are two types of parsing errors to a token in a dependency parse tree. One is that the token points to a wrong head, namely dependenterror, and another one is that the token is recognized wrongly as a head of other tokens, namely head-error. For example, Figure 3 presents a possible wrong parse tree of the example shown in Figure 1 . By comparing with the gold-tree in Figure 1 , tokens (POS tag) of \"he (PN)\", \"went (VV)\", \"bookstore (NN)\", \"buy (VV)\", \"a (CD)\", and \". (PU)\" in the dependency tree in Figure 3 all point to different wrong heads, which are dependent-errors. Concurrently, tokens (POS tag) of \"went (VV)\", \"buy (VV)\", and \"book (NN)\" are wrongly recognized as heads of other tokens (e.g., \"he\", \"bookstore\", \"a\"), which are head-errors. According to the definition, every head-error has at least one corresponding dependent-error. However, in the case that a token is not the root in a gold-tree but is root in the wrong tree, this token is a dependent-error corresponding with no headerror. An example is the dependent-error \"went (VV)\" in Figure 3 .", "cite_spans": [], "ref_spans": [ { "start": 261, "end": 269, "text": "Figure 3", "ref_id": null }, { "start": 331, "end": 339, "text": "Figure 1", "ref_id": null }, { "start": 377, "end": 385, "text": "Figure 1", "ref_id": null }, { "start": 511, "end": 517, "text": "Figure", "ref_id": null }, { "start": 1066, "end": 1074, "text": "Figure 3", "ref_id": null } ], "eq_spans": [], "section": "Part-of-Speech Tag Error", "sec_num": "5.1" }, { "text": "We count the number of POS tag misrecognitions separately for dependent-and headerrors. In the example of Figure 3 , dependenterror counts are for VV, 2 errors, and PN, NN, CD, PU each 1 error. The number of POS tag mis-recognitions for head-errors are VV with 2 errors, and NN with 1 error. In our analysis, we will compute these counts for all POS tags at every sentence in our data set. However, our reordering method performed differently at each sentence in our data set, and the reordering quality varied from sentence to sentence. With the objective of observing the correlation between reordering quality and each type of error, we will first group sentences according to their Kendall's \u03c4 values. Then, we will compute proportions of POS tag errors at each \u03c4 value, for every type of POS tag error. Figure 4 shows the distribution of top three dependent-error POS tags, which means that they are the three most frequent POS tags that point to a wrong head in auto-parse trees. VV represents all verbs except predicative adjective (VA), copula (VC), and you3 7 as the main verb (VE). PU represents punctuation and NN represents all nouns except proper noun (NR), temporal noun (NT), and the ones for locations which cannot modify verb phrases with or without de0 8 . The dependenterror on VV accounts for a larger proportion in low reordering accuracy sentences whereas more NN dependent-error occurred in high reordering accuracy sentences. On the other hand, the proportion of PU dependent-error is more consistent. Figure 5 shows the distribution of top two headerror POS tags, which means that they are the two most frequent POS tags that are recognized wrongly as heads in auto-parse trees. Comparing to Figure 4 , the tendency of both VV and NN is the same but distincter.", "cite_spans": [], "ref_spans": [ { "start": 106, "end": 114, "text": "Figure 3", "ref_id": null }, { "start": 808, "end": 816, "text": "Figure 4", "ref_id": "FIGREF1" }, { "start": 1526, "end": 1534, "text": "Figure 5", "ref_id": "FIGREF2" }, { "start": 1717, "end": 1725, "text": "Figure 4", "ref_id": "FIGREF1" } ], "eq_spans": [], "section": "Part-of-Speech Tag Error", "sec_num": "5.1" }, { "text": "The analysis results on the proportion distributions of dependent-error POS tags and head-error POS tags in different reordering quality sentence groups exhibit that there are more parsing errors on verbs than nouns in low reordering accuracy sentences and thus the parsing errors on verbs influence more on the reordering performance. However, it is still difficult to reveal the effects of more concrete parsing errors on reordering consid- ering that not all verb parsing errors influence the reordering. As an illustration, in Figure 3 , if the head of \"bookstore\" were \"went\", the VV headerror of \"went\" would not cause any reordering error since it would be reordered consistently to the right-hand side of its RM-D \"bookstore\". Consequently, we use a descriptive approach to analyze dependency types to explore the effects from more concrete parsing errors in the next section.", "cite_spans": [], "ref_spans": [ { "start": 531, "end": 539, "text": "Figure 3", "ref_id": null } ], "eq_spans": [], "section": "Part-of-Speech Tag Error", "sec_num": "5.1" }, { "text": "As introduced in Section 2.1, DPC first identifies Vb, RM-D, and then reorders necessary words. Thus, DPC reorders not only Vb-H, but also Vb-D in a Vb, which means that the failure on identifying Vbs may also cause unexpected reordering on particles, such as aspect markers. However, in this work, we only focus on reordering issues of Vb-H candidates 9 . To discover the effects of more concrete parsing errors on reordering, we distinguish three categories of dependency types, i.e., ROOT, RM-D, and BEI. Among them, ROOT denotes whether the Vb-H candidate is the root of the sentence or not, RM-D is the right-most object dependent of the Vb-H candidate if it has one, and BEI denotes whether the Vb-H candidate is involved in a bei-construction. According to the methodology of the reordering method DPC, we define seven patterns of parsing error phenomena and classify them into three types by comparing the gold-tree (GT) with autoparse tree (Corbit-tree, CT). Table 4 lists all parsing error patterns in three error types, ROOT error, RM-D error and BEI error by considering three dependency types ROOT, RM-D and BEI. Symbols of \" \u221a \", \"\u00d7\", \"?\" represent the status of a cer- ", "cite_spans": [], "ref_spans": [ { "start": 968, "end": 975, "text": "Table 4", "ref_id": "TABREF5" } ], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": "BEI ROOT RM-D GT CT GT CT GT CT ROOT Error Root-C \u00d7 \u00d7 \u00d7 \u221a \u00d7 \u00d7 Root-G \u00d7 \u00d7 \u221a \u00d7 \u00d7 \u00d7 RM-D Error RM D-C \u00d7 \u00d7 \u00d7 \u00d7 \u00d7 \u221a \u00d7 \u00d7 \u00d7 \u221a \u00d7 \u221a \u00d7 \u00d7 \u221a \u00d7 \u00d7 \u221a \u00d7 \u00d7 \u221a \u221a \u00d7 \u221a RM D-G \u00d7 \u00d7 \u00d7 \u00d7 \u221a \u00d7 \u00d7 \u00d7 \u00d7 \u221a \u221a \u00d7 \u00d7 \u00d7 \u221a \u00d7 \u221a \u00d7 \u00d7 \u00d7 \u221a \u221a \u221a \u00d7 RM D-D \u00d7 \u00d7 \u00d7 \u00d7 \u221a diff. \u00d7 \u00d7 \u00d7 \u221a \u221a diff. \u00d7 \u00d7 \u221a \u00d7 \u221a diff. \u00d7 \u00d7 \u221a \u221a \u221a diff. BEI Error BEI-C \u00d7 \u221a \u221a ? \u00d7 ? \u00d7 \u221a \u00d7 ? \u221a ? \u00d7 \u221a \u221a ? \u221a ? BEI-G \u221a \u00d7 ? \u221a ? \u00d7 \u221a \u00d7 ? \u00d7 ? \u221a \u221a \u00d7 ? \u221a ? \u221a", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": "(Root-C, Root- G, RM D-C, RM D-G, RM D-D, BEI-C, BEI-G)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": "that cause three types of reordering issues (ROOT error, RM-D error, and BEI error). GT stands for gold-tree, and CT stands for Corbit-tree. Symbols \" \u221a \", \"\u00d7\", \"?\" represent the status of True, False, and Unknown, respectively. \"diff.\" means that the RM-Ds exist in both GT and CT but are different. tain dependency type in gold-tree or Corbit-tree. For every Vb-H candidate, the 6 status are conditions to match the error pattern. For example, to match a Root-C error pattern, the Vb-H candidate needs to satisfy the following conditions: in goldtree, it is not the root, and does not have any RM-D or bei dependent; in Corbit tree, it does not have any RM-D or bei dependent, but it is the root.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": "Root-C is the case where a Vb-H candidate has been wrongly parsed as the root of the sentence. However, it only affects the reordering with two constrains, namely that RM-D of the Vb-H candidate does not exist and Vb-H is not involved in a bei-construction. For instance, the Vb-H \"should\" in the example of Figure 6 was recognized as root in auto-parse tree in Figure 6b . However, the actual root is the Vb-H \"is\" in gold tree of Figure 6a . Therefore, since \"should\" does not have any dependent as either BEI or RM-D in both GT and CT, it will be reordered incorrectly to the end of ", "cite_spans": [], "ref_spans": [ { "start": 308, "end": 316, "text": "Figure 6", "ref_id": "FIGREF4" }, { "start": 362, "end": 371, "text": "Figure 6b", "ref_id": "FIGREF4" }, { "start": 432, "end": 441, "text": "Figure 6a", "ref_id": "FIGREF4" } ], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": ". ROOT . . o . . o . . o . . o . . o . . o . . o . . o . . o . . o .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Dependency Type Error", "sec_num": "5.2" }, { "text": ".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u7d50\u8ad6\u7684 (\u306b)", "sec_num": null }, { "text": ".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u8a00\u3046 (\u3068)", "sec_num": null }, { "text": ".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u8ecd\u8266 (\u304c)", "sec_num": null }, { "text": ".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u6d77\u8ecd (\u306b)", "sec_num": null }, { "text": ".", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u52a0\u308f\u3063\u3066", "sec_num": null }, { "text": ". One should say that, the additions of warships will help to improve the navy's combat power.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u6226\u529b (\u306f)", "sec_num": null }, { "text": "the sentence according to the CT whereas it will not be reordered according to GT, which is already in the same position as its Japanese counterpart. Root-G is the opposite case of Root-C where a Vb-H candidate is the root of the sentence but was not parsed as the root in CT. This affects the reordering under the two same constraints as Root-C. Figure 7b shows an example of Root-G. In Figure 7a, the word alignment shows that the Vb-H \"agree\" should be reordered to the end of the sentence. However, it will not be reordered for the wrong parse tree shown in Figure 7b .", "cite_spans": [], "ref_spans": [ { "start": 347, "end": 356, "text": "Figure 7b", "ref_id": "FIGREF6" }, { "start": 388, "end": 394, "text": "Figure", "ref_id": null }, { "start": 562, "end": 571, "text": "Figure 7b", "ref_id": "FIGREF6" } ], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "RM D-C is the case where the RM-D of a Vb-H candidate exists in a CT but not in GT. In other words, a RM-D candidate was parsed wrongly on its head. There are four varieties of combination with the status of ROOT, BEI of the Vb-H candidate that lead to incorrect reorderings. The Vb-H \"agree\" in Figure 7c matches the last combination of RM D-C, which will be reordered right after \"journalist\" instead of at the end of the sentence.", "cite_spans": [], "ref_spans": [ { "start": 296, "end": 305, "text": "Figure 7c", "ref_id": "FIGREF6" } ], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "RM D-G is the opposite case of RM D-C where the RM-D of a Vb-H candidate was missed in a CT. There are also four cases of reordering errors according to the status of BEI, ROOT and RM-D. Vb-H \"went\" in Figure 3 matches the second combination of RM D-G so that it will not be able to reorder after \"bookstore\".", "cite_spans": [], "ref_spans": [ { "start": 202, "end": 210, "text": "Figure 3", "ref_id": null } ], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "RM D-D is the case where a bei-constructionfree Vb-H candidate obtains two different RM-D candidates in CT and GT, which causes the reordering issue. In Figure 6 , Vb-H \"join\" received different RM-Ds in two trees. word alignment, it should be reordered next to \"navy\" instead of \"combat power\". BEI-C is the case where a Vb-H candidate received a wrong BEI dependent in CT. This will prevent reordering independently on whether the Vb-H candidate has RM-D or is the root.", "cite_spans": [], "ref_spans": [ { "start": 153, "end": 161, "text": "Figure 6", "ref_id": "FIGREF4" } ], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "BEI-G is the opposite case of BEI-C, where Vb-H in GT will not be reordered but in CT it will.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "After defining seven patterns of parsing errors and classifying them into three types, we calculate the average frequency proportions of each type in different \u03c4 value groups of sentences. Figure 8 shows the distribution of the three types of parsing errors and their tendencies. In low \u03c4 value sentences, there are higher proportions of ROOT errors, and relatively lower proportions in high \u03c4 value sentences. RM-D errors follow the opposite tendency. This implies that the effects of ROOT errors on reordering are stronger than the effects from RM-D errors. The reason could be that ROOT errors cause long distance reordering failure while RM-D errors lead to more local reordering errors. Since there are very few BEI errors, it was difficult to capture their trends. Figure 9 and Figure 10 provide the correlations between parsing error patterns and reordering accuracy. In ROOT errors types, Root-C had a larger percentage than Root-G in low reordering accuracy sentences which shows that the Vb-H can- didate that does not have any object dependent tends to be recognized as root by parser. This is consistent with the distribution results that are shown in Figure 10 . The error pattern of RM D-G had larger percentage than the other two patterns, which also implies that a Vb-H candidate in a CT tends to have less or none object dependents.", "cite_spans": [], "ref_spans": [ { "start": 189, "end": 197, "text": "Figure 8", "ref_id": "FIGREF7" }, { "start": 771, "end": 779, "text": "Figure 9", "ref_id": "FIGREF8" }, { "start": 784, "end": 793, "text": "Figure 10", "ref_id": "FIGREF9" }, { "start": 1164, "end": 1173, "text": "Figure 10", "ref_id": "FIGREF9" } ], "eq_spans": [], "section": "\u5411\u4e0a", "sec_num": null }, { "text": "Due to the time limitation, we only focused on analyzing parsing errors that cause reordering issues on Vb-H candidates while defining the error patterns. However, it is not only that Vb-H candidates are reordered in DPC, but also other words like Vb-D candidates and particles will be reordered. It is also meaningful to explore the parsing error patterns which cause unexpected reordering on these words and the correlation between them as well.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Further Analysis Possibilities", "sec_num": "5.3" }, { "text": "The current study on exploring influential parsing errors is not exhaustive, and another analysis possibility would be to explore what types of parsing errors do not affect reordering so that parsers can sacrifice their performance on those types of issues in order to improve on influential types. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Further Analysis Possibilities", "sec_num": "5.3" }, { "text": "Two important research directions concentrate on either improving parsers or developing linguistically motivated pre-reordering methods. We believe that analyzing the link between those directions can help us to refine future developments.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "We observed relatively small effects on reordering quality in response of parsing errors. However, reordering quality affect word alignments, which in turn affect the quality of bilingual phrases that are extracted. It would be interesting to extend this work to quantify the propagation of parsing and reordering errors in SMT pipelines, to observe the factored effect on the overall MT quality.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "We found that not all POS tagging and parsing errors correlate equally with reordering quality. In the case of DPC reordering method, misrecognitions of VV words correlate with low reordering performance, whereas mis-recognitions of NN words had a smaller impact. Indeed, DPC heavily relies on detecting verbal blocks that are candidates for reordering, and systems that use the same strategy should choose POS taggers that display high accuracy of VV recognition.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "One of the key characteristics of DPC is its ability to correctly reorder sentences with reported speech constructions. For that purpose, it is crucial for parsers to recognize the sentence root, and our analysis demonstrated that systems that follow similar strategy should rely on parsers that have a high accuracy to recognize the sentence root.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "In general, we believe that future developments of syntax-based pre-reordering methods would benefit of preliminary analysis of POS tagging and parsing accuracies. In case of linguistically motivated pre-reordering methods, reordering rules could be designed to be more robust against unreliable POS tags or unreliable dependency relations. For automatically learned reordering rules, those systems could be designed to make use of N-best lists of certain POS tags or dependencies that are critical but that parsers cannot reliably provide.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "There are other popular syntax-based prereordering methods that may use different types of parsing grammars (i.e. Head-driven phrase structure grammar), and similar analysis would also be interesting in those contexts, possibly with a larger set of gold parsed and reordered sentences. Additionally, researchers interested in developing POS taggers and parsers with the objective to aid pre-reordering could attempt to maximize the accuracy of POS tags or dependencies that are relevant to the reordering task, maybe at the expense of lower accuracies on other elements.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Discussion and Future Work", "sec_num": "6" }, { "text": "In this work, we carried out linguistically motivated analysis methods by combining empirical and descriptive approaches in three analysis stages to examine the effects of different parsing errors on pre-reordering performance. We achieved four objectives: (i) quantify effects of parsing errors on reordering, (ii) estimate upper bounds in performance of the reordering method, (iii) profile general parsing errors, and (iv) examine effects of specific parsing errors on reordering.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "7" }, { "text": "In the first stage, we set up benchmarks in two scenarios for reordered Chinese sentences. By calculating the word order similarity between the benchmarks and the dependency parse tree based auto-reordered Chinese sentences, we quantified the correlation between parsing errors and reordering accuracies as well as explored the upper bound in reordering quality of the reordering model.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "7" }, { "text": "In the second stage, we examined the effects of two types of parsing errors on reordering quality by using POS tag information. The distributions of parsing errors' POS tags provide a general view of the influential parsing error types and an approximation to the cause of the effects.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "7" }, { "text": "In the last stage, we defined several patterns of parsing errors that assuredly cause reordering errors by using the linguistic feature of dependency types based on a deep linguistic study of the syntactic structures and the reordering model. The analysis results assist us to achieve a better and more explicit understanding on the relationship between parsing errors and reordering performance. Furthermore, we captured the effects of more concrete parsing errors on reordering.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "7" }, { "text": "In this work, POS tag definitions follow the POS tag guidelines of the Penn Chinese Treebank v3.0.2 According to(Han et al., 2013), a Vb includes the head of the Vb (Vb-H) and an optional component (Vb-D).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "http://triplet.cc/software/corbit 5 Note that Corbit was tuned with the development set of CTB-7.PACLIC-27", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "A Chinese character expresses possession and existence. 8 A Chinese character is specially used to connect the verb phrase and its modifier.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "We use \"Vb-H candidate\" in this work for the reason that if the Vb-H is involved into a bei-construction, then it can not be Vb-H according to(Han et al., 2013).PACLIC-27", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "Frustratingly hard domain adaptation for dependency parsing", "authors": [ { "first": "Mark", "middle": [], "last": "Dredze", "suffix": "" }, { "first": "John", "middle": [], "last": "Blitzer", "suffix": "" }, { "first": "Partha", "middle": [ "Pratim" ], "last": "Talukdar", "suffix": "" }, { "first": "Kuzman", "middle": [], "last": "Ganchev", "suffix": "" }, { "first": "Joao", "middle": [], "last": "Graca", "suffix": "" }, { "first": "Fernando", "middle": [], "last": "Pereira", "suffix": "" } ], "year": 2007, "venue": "Proc. of the CoNLL Shared Task Session of EMNLP-CoNLL", "volume": "", "issue": "", "pages": "1051--1055", "other_ids": {}, "num": null, "urls": [], "raw_text": "Mark Dredze, John Blitzer, Partha Pratim Taluk- dar, Kuzman Ganchev, Joao Graca, and Fer- nando Pereira. 2007. Frustratingly hard do- main adaptation for dependency parsing. In Proc. of the CoNLL Shared Task Session of EMNLP-CoNLL, pages 1051-1055.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Parallel implementations of word alignment tool", "authors": [ { "first": "Qin", "middle": [], "last": "Gao", "suffix": "" }, { "first": "Stephan", "middle": [], "last": "Vogel", "suffix": "" } ], "year": 2008, "venue": "Software Engineering, Testing, and Quality Assurance for Natural Language Processing", "volume": "", "issue": "", "pages": "49--57", "other_ids": {}, "num": null, "urls": [], "raw_text": "Qin Gao and Stephan Vogel. 2008. Parallel im- plementations of word alignment tool. In Soft- ware Engineering, Testing, and Quality Assur- ance for Natural Language Processing, pages 49-57.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Automatically learning source-side reordering rules for large scale machine translation", "authors": [ { "first": "Dmitriy", "middle": [], "last": "Genzel", "suffix": "" } ], "year": 2010, "venue": "Proc. of COLING", "volume": "", "issue": "", "pages": "376--384", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dmitriy Genzel. 2010. Automatically learning source-side reordering rules for large scale ma- chine translation. In Proc. of COLING, pages 376-384.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Towards heterogeneous automatic MT error analysis", "authors": [ { "first": "Jes\u00fas", "middle": [], "last": "Gim\u00e9nez", "suffix": "" }, { "first": "Lluis", "middle": [], "last": "M\u00e0rquez", "suffix": "" } ], "year": 2008, "venue": "Proc. of the 6th International Conference on Language Resources and Evaluation", "volume": "", "issue": "", "pages": "1894--1901", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jes\u00fas Gim\u00e9nez and Lluis M\u00e0rquez. 2008. To- wards heterogeneous automatic MT error analy- sis. In Proc. of the 6th International Conference on Language Resources and Evaluation, pages 1894-1901.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Effects of noun phrase bracketing in dependency parsing and machine translation", "authors": [ { "first": "Nathan", "middle": [], "last": "Green", "suffix": "" } ], "year": 2011, "venue": "Proc. of ACL-HLT, Student Session", "volume": "", "issue": "", "pages": "69--74", "other_ids": {}, "num": null, "urls": [], "raw_text": "Nathan Green. 2011. Effects of noun phrase bracketing in dependency parsing and machine translation. In Proc. of ACL-HLT, Student Ses- sion, pages 69-74.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Head finalization reordering for Chineseto-Japanese machine translation", "authors": [ { "first": "Dan", "middle": [], "last": "Han", "suffix": "" }, { "first": "Katsuhito", "middle": [], "last": "Sudoh", "suffix": "" }, { "first": "Xianchao", "middle": [], "last": "Wu", "suffix": "" }, { "first": "Kevin", "middle": [], "last": "Duh", "suffix": "" }, { "first": "Hajime", "middle": [], "last": "Tsukada", "suffix": "" }, { "first": "Masaaki", "middle": [], "last": "Nagata", "suffix": "" } ], "year": 2012, "venue": "Proc. of the ACL 6th Workshop on SSST", "volume": "", "issue": "", "pages": "57--66", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dan Han, Katsuhito Sudoh, Xianchao Wu, Kevin Duh, Hajime Tsukada, and Masaaki Nagata. 2012. Head finalization reordering for Chinese- to-Japanese machine translation. In Proc. of the ACL 6th Workshop on SSST, pages 57-66.", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "Using unlabeled dependency parsing for pre-reordering for Chinese-to-Japanese statistical machine translation", "authors": [ { "first": "Dan", "middle": [], "last": "Han", "suffix": "" }, { "first": "Yusuke", "middle": [], "last": "Mart\u00ednez-G\u00f3mez", "suffix": "" }, { "first": "Katsuhito", "middle": [], "last": "Miyao", "suffix": "" }, { "first": "Masaaki", "middle": [], "last": "Sudoh", "suffix": "" }, { "first": "", "middle": [], "last": "Nagata", "suffix": "" } ], "year": 2013, "venue": "Proc. of the ACL Second Workshop on Hybrid Approaches to Translation", "volume": "", "issue": "", "pages": "25--33", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dan Han, Pascual Mart\u00ednez-G\u00f3mez, Yusuke Miyao, Katsuhito Sudoh, and Masaaki Nagata. 2013. Using unlabeled dependency parsing for pre-reordering for Chinese-to-Japanese sta- tistical machine translation. In Proc. of the ACL Second Workshop on Hybrid Approaches to Translation, pages 25-33.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Descriptive and empirical approaches to capturing underlying dependencies among parsing errors", "authors": [ { "first": "Tadayoshi", "middle": [], "last": "Hara", "suffix": "" }, { "first": "Yusuke", "middle": [], "last": "Miyao", "suffix": "" }, { "first": "Jun'ichi", "middle": [], "last": "Tsujii", "suffix": "" } ], "year": 2009, "venue": "Proc. of EMNLP", "volume": "", "issue": "", "pages": "1162--1171", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tadayoshi Hara, Yusuke Miyao, and Jun'ichi Tsu- jii. 2009. Descriptive and empirical approaches to capturing underlying dependencies among parsing errors. In Proc. of EMNLP, pages 1162-1171.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "Incremental joint POS tagging and dependency parsing in Chinese", "authors": [ { "first": "Jun", "middle": [], "last": "Hatori", "suffix": "" }, { "first": "Takuya", "middle": [], "last": "Matsuzaki", "suffix": "" }, { "first": "Yusuke", "middle": [], "last": "Miyao", "suffix": "" }, { "first": "Junichi", "middle": [], "last": "Tsujii", "suffix": "" } ], "year": 2011, "venue": "Proc. of 5th IJCNLP", "volume": "", "issue": "", "pages": "1216--1224", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jun Hatori, Takuya Matsuzaki, Yusuke Miyao, and Junichi Tsujii. 2011. Incremental joint POS tagging and dependency parsing in Chinese. In Proc. of 5th IJCNLP, pages 1216-1224.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "Head finalization: A simple reordering rule for SOV languages", "authors": [ { "first": "Hideki", "middle": [], "last": "Isozaki", "suffix": "" }, { "first": "Katsuhito", "middle": [], "last": "Sudoh", "suffix": "" }, { "first": "Hajime", "middle": [], "last": "Tsukada", "suffix": "" }, { "first": "Kevin", "middle": [], "last": "Duh", "suffix": "" } ], "year": 2010, "venue": "Proc. of WMTMetricsMATR", "volume": "", "issue": "", "pages": "244--251", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hideki Isozaki, Katsuhito Sudoh, Hajime Tsukada, and Kevin Duh. 2010. Head final- ization: A simple reordering rule for SOV languages. In Proc. of WMTMetricsMATR, pages 244-251.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Training a parser for machine translation reordering", "authors": [ { "first": "Jason", "middle": [], "last": "Katz-Brown", "suffix": "" }, { "first": "Slav", "middle": [], "last": "Petrov", "suffix": "" }, { "first": "Ryan", "middle": [], "last": "Mcdonald", "suffix": "" }, { "first": "Franz", "middle": [], "last": "Och", "suffix": "" }, { "first": "David", "middle": [], "last": "Talbot", "suffix": "" }, { "first": "Hiroshi", "middle": [], "last": "Ichikawa", "suffix": "" }, { "first": "Masakazu", "middle": [], "last": "Seno", "suffix": "" } ], "year": 2011, "venue": "Proc. of the 2011 Conference on EMNLP", "volume": "", "issue": "", "pages": "183--192", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jason Katz-Brown, Slav Petrov, Ryan McDonald, Franz Och, David Talbot, Hiroshi Ichikawa, and Masakazu Seno. 2011. Training a parser for machine translation reordering. In Proc. of the 2011 Conference on EMNLP, pages 183-192.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "Japanese dependency structure analysis based on support vector machines", "authors": [ { "first": "Taku", "middle": [], "last": "Kudo", "suffix": "" }, { "first": "Yuji", "middle": [], "last": "Matsumoto", "suffix": "" } ], "year": 2000, "venue": "Proc. of the EMNLP/VLC-2000", "volume": "", "issue": "", "pages": "18--25", "other_ids": {}, "num": null, "urls": [], "raw_text": "Taku Kudo and Yuji Matsumoto. 2000. Japanese dependency structure analysis based on support vector machines. In Proc. of the EMNLP/VLC- 2000, pages 18-25.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "Characterizing the errors of data-driven dependency parsing models", "authors": [ { "first": "Ryan", "middle": [], "last": "Mcdonald", "suffix": "" }, { "first": "Joakim", "middle": [], "last": "Nivre", "suffix": "" } ], "year": 2007, "venue": "Proc. of the 2007 Joint Conference on EMNLP-CoNLL", "volume": "", "issue": "", "pages": "122--131", "other_ids": {}, "num": null, "urls": [], "raw_text": "Ryan McDonald and Joakim Nivre. 2007. Char- acterizing the errors of data-driven dependency parsing models. In Proc. of the 2007 Joint Con- ference on EMNLP-CoNLL, pages 122-131.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Feature forest models for probabilistic HPSG parsing", "authors": [ { "first": "Yusuke", "middle": [], "last": "Miyao", "suffix": "" }, { "first": "Jun'ichi", "middle": [], "last": "Tsujii", "suffix": "" } ], "year": 2008, "venue": "Computational Linguistics", "volume": "34", "issue": "", "pages": "35--80", "other_ids": {}, "num": null, "urls": [], "raw_text": "Yusuke Miyao and Jun'ichi Tsujii. 2008. Feature forest models for probabilistic HPSG parsing. Computational Linguistics, 34:35-80.", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "Headdriven phrase structure grammar. The University of", "authors": [ { "first": "Jesse", "middle": [], "last": "Carl", "suffix": "" }, { "first": "Ivan", "middle": [ "A" ], "last": "Pollard", "suffix": "" }, { "first": "", "middle": [], "last": "Sag", "suffix": "" } ], "year": 1994, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Carl Jesse Pollard and Ivan A. Sag. 1994. Head- driven phrase structure grammar. The Univer- sity of Chicago Press and CSLI Publications.", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "The impact of parse quality on syntacticallyinformed statistical machine translation", "authors": [ { "first": "Chris", "middle": [], "last": "Quirk", "suffix": "" }, { "first": "Simon", "middle": [], "last": "Corston-Oliver", "suffix": "" } ], "year": 2006, "venue": "Proc. of EMNLP", "volume": "", "issue": "", "pages": "62--69", "other_ids": {}, "num": null, "urls": [], "raw_text": "Chris Quirk and Simon Corston-Oliver. 2006. The impact of parse quality on syntactically- informed statistical machine translation. In Proc. of EMNLP, pages 62-69.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "Improving Chinese word segmentation and POS tagging with semisupervised methods using large auto-analyzed data", "authors": [ { "first": "Yiou", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Junichi", "middle": [], "last": "Kazama", "suffix": "" }, { "first": "Yoshimasa", "middle": [], "last": "Tsuruoka", "suffix": "" }, { "first": "Wenliang", "middle": [], "last": "Chen", "suffix": "" }, { "first": "Yujie", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "Kentaro", "middle": [], "last": "Torisawa", "suffix": "" } ], "year": 2011, "venue": "Proc. of 5th IJCNLP", "volume": "", "issue": "", "pages": "309--317", "other_ids": {}, "num": null, "urls": [], "raw_text": "Yiou Wang, Junichi Kazama, Yoshimasa Tsu- ruoka, Wenliang Chen, Yujie Zhang, and Ken- taro Torisawa. 2011. Improving Chinese word segmentation and POS tagging with semi- supervised methods using large auto-analyzed data. In Proc. of 5th IJCNLP, pages 309-317.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Improving a statistical MT system with automatically learned rewrite patterns", "authors": [ { "first": "Fei", "middle": [], "last": "Xia", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Mccord", "suffix": "" } ], "year": 2004, "venue": "Proc. of COLING", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Fei Xia and Michael McCord. 2004. Improv- ing a statistical MT system with automatically learned rewrite patterns. In Proc. of COLING.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Using a dependency parser to improve SMT for subject-object-verb languages", "authors": [ { "first": "Peng", "middle": [], "last": "Xu", "suffix": "" }, { "first": "Jaeho", "middle": [], "last": "Kang", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Ringgaard", "suffix": "" }, { "first": "Franz", "middle": [], "last": "Och", "suffix": "" } ], "year": 2009, "venue": "Proc. of NAACL", "volume": "", "issue": "", "pages": "245--253", "other_ids": {}, "num": null, "urls": [], "raw_text": "Peng Xu, Jaeho Kang, Michael Ringgaard, and Franz Och. 2009. Using a dependency parser to improve SMT for subject-object-verb lan- guages. In Proc. of NAACL, pages 245-253.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Analysis of the difficulties in Chinese deep parsing", "authors": [ { "first": "Kun", "middle": [], "last": "Yu", "suffix": "" }, { "first": "Yusuke", "middle": [], "last": "Miyao", "suffix": "" }, { "first": "Takuya", "middle": [], "last": "Matsuzaki", "suffix": "" }, { "first": "Xiangli", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Junichi", "middle": [], "last": "Tsujii", "suffix": "" } ], "year": 2011, "venue": "Proc. of the 12th International Conference on Parsing Technologies", "volume": "", "issue": "", "pages": "48--57", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kun Yu, Yusuke Miyao, Takuya Matsuzaki, Xian- gli Wang, and Junichi Tsujii. 2011. Analysis of the difficulties in Chinese deep parsing. In Proc. of the 12th International Conference on Parsing Technologies, pages 48-57.", "links": null } }, "ref_entries": { "FIGREF0": { "uris": null, "num": null, "type_str": "figure", "text": "The distribution of Kendall's tau values for 2, 236 bilingual sentences (Chinese-Japanese) in which the Chinese is from three systems of baseline, Auto-DPC, and Gold-DPC." }, "FIGREF1": { "uris": null, "num": null, "type_str": "figure", "text": "The distribution of top three dependenterror POS tags and their tendency lines." }, "FIGREF2": { "uris": null, "num": null, "type_str": "figure", "text": "The distribution of top two head-error POS tags and their tendency lines." }, "FIGREF4": { "uris": null, "num": null, "type_str": "figure", "text": "An example for parsing error patterns of Root-C and RM D-D. English translation:" }, "FIGREF5": { "uris": null, "num": null, "type_str": "figure", "text": "Another possible erroneous parse tree." }, "FIGREF6": { "uris": null, "num": null, "type_str": "figure", "text": "An example for parsing error patterns of Root-G and RM D-C. English translation: He agreed to the journalist to take a picture of him." }, "FIGREF7": { "uris": null, "num": null, "type_str": "figure", "text": "Distribution of three types of parsing errors in different \u03c4 groups and their trend curves. Mov. Avg. (ROOT-G) 2 per. Mov. Avg. (ROOT-C)" }, "FIGREF8": { "uris": null, "num": null, "type_str": "figure", "text": "Distribution of patterns of ROOT error in different \u03c4 groups and their trend curves." }, "FIGREF9": { "uris": null, "num": null, "type_str": "figure", "text": "The distribution of different patterns of RM-D error in different \u03c4 groups." }, "TABREF1": { "content": "", "text": "Statistics of selected sentences in five genres of CTB-7. AL stands for the average length of sentences, while Voc. for vocabulary.", "num": null, "html": null, "type_str": "table" }, "TABREF2": { "content": "
Baseline Gold-DPC Auto-DPC
M-reordered0.820.900.88
Gold-DPC--0.95
", "text": "The set of Japanese references plays the role of benchmark and is compared with the set of automatically reordered Chi-3 http://stp.lingfil.uu.se/ nivre/research/Penn2Malt.html", "num": null, "html": null, "type_str": "table" }, "TABREF3": { "content": "", "text": "The average value of Kendall's tau (\u03c4 ) of 517 Chinese sentences by comparing manually reordered sentences, unreordered sentences, and automatically reordered sentences. M-reordered is short for manually reordered. nese sentences. Word alignments are produced by MGIZA++", "num": null, "html": null, "type_str": "table" }, "TABREF5": { "content": "
", "text": "", "num": null, "html": null, "type_str": "table" } } } }