|
{ |
|
"paper_id": "1994", |
|
"header": { |
|
"generated_with": "S2ORC 1.0.0", |
|
"date_generated": "2023-01-19T02:06:23.842514Z" |
|
}, |
|
"title": "LEARNING AN ENGLISH-CHINESE LEXICON FROM A PARALLEL CORPUS", |
|
"authors": [ |
|
{ |
|
"first": "Dekai", |
|
"middle": [], |
|
"last": "Wu", |
|
"suffix": "", |
|
"affiliation": { |
|
"laboratory": "", |
|
"institution": "University of Science & Technology Clear Water Bay", |
|
"location": { |
|
"settlement": "Hong Kong" |
|
} |
|
}, |
|
"email": "" |
|
}, |
|
{ |
|
"first": "Xuanyin", |
|
"middle": [], |
|
"last": "Xia", |
|
"suffix": "", |
|
"affiliation": { |
|
"laboratory": "", |
|
"institution": "University of Science & Technology Clear Water Bay", |
|
"location": { |
|
"settlement": "Hong Kong" |
|
} |
|
}, |
|
"email": "" |
|
} |
|
], |
|
"year": "", |
|
"venue": null, |
|
"identifiers": {}, |
|
"abstract": "We report experiments on automatic learning of an English-Chinese translation lexicon, through statistical training on a large parallel corpus. The learned vocabulary size is nontrivial at 6,517 English words averaging 2.33 Chinese translations per entry, with a manuallyfiltered precision of 95.1% and a single-most-probable precision of 91.2%. We then introduce a significance filtering method that is fully automatic, yet still yields a weighted precision of 86.0%. Learning of translations is adaptive to the domain. To our knowledge, these are the first empirical results of the kind between an Indo-European and non-Indo-European language for any significant corpus size with a non-toy vocabulary.", |
|
"pdf_parse": { |
|
"paper_id": "1994", |
|
"_pdf_hash": "", |
|
"abstract": [ |
|
{ |
|
"text": "We report experiments on automatic learning of an English-Chinese translation lexicon, through statistical training on a large parallel corpus. The learned vocabulary size is nontrivial at 6,517 English words averaging 2.33 Chinese translations per entry, with a manuallyfiltered precision of 95.1% and a single-most-probable precision of 91.2%. We then introduce a significance filtering method that is fully automatic, yet still yields a weighted precision of 86.0%. Learning of translations is adaptive to the domain. To our knowledge, these are the first empirical results of the kind between an Indo-European and non-Indo-European language for any significant corpus size with a non-toy vocabulary.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Abstract", |
|
"sec_num": null |
|
} |
|
], |
|
"body_text": [ |
|
{ |
|
"text": "A criticism of statistical machine translation tools is that convincing empirical results to date are largely confined to similar language pairs, such as French and English. We offer some contributions to the pool of evidence supporting the language-independence of statistical techniques, from the SILC project at HKUST which is studying machine learning of natural language translation. Specifically, we report accuracy rates for a new performance measure: the precision of statistical translation lexicon acquisition, between English and Chinese.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Introduction", |
|
"sec_num": "1" |
|
}, |
|
{ |
|
"text": "In the first phase of SILC (statistical inter-lingual conversion), we have (1) collected a bilingual corpus of parallel English and Chinese text, (2) aligned the sentences within the corpus, and (3) learned a bilingual lexicon from the aligned data, to be embedded within learned translation models. Preliminary results on the first two steps were reported in Wu (1994) ; an updated summary is given in section 2, before we discuss the bilingual lexicon learning.", |
|
"cite_spans": [ |
|
{ |
|
"start": 360, |
|
"end": 369, |
|
"text": "Wu (1994)", |
|
"ref_id": "BIBREF11" |
|
} |
|
], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Introduction", |
|
"sec_num": "1" |
|
}, |
|
{ |
|
"text": "One motivation for this work is to conduct English-Chinese \"acid tests\" of statistical NLP techniques. Another benefit of the approach is that it obtains not only a translation lexicon, but also the probabilities of alternative translations for the same word. A further advantage of the learning approach is the ability to acquire lexicons that are adapted for particular domains or genres. The vocabulary of our corpus, for example, includes a high proportion of words not found in the English-Chinese machine-readable dictionaries we have seen.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Introduction", |
|
"sec_num": "1" |
|
}, |
|
{ |
|
"text": "Though large parallel bilingual corpora are relatively scarce compared with monolingual corpora, they have generated more interesting results. Significant progress has been made on problems including automatic sentence alignment (Kay & R\u00f6scheisen 1988; Catizone et al. 1989; Gale & Church 1991; Brown et al. 1991; Chen 1993) , coarse alignment Fung & Church 1994) , statistical machine translation (Brown et al. 1990; Brown et al. 1993) , word alignment (Dagan et al. 1993) , word sense disambiguation (Gale et al. 1993) , and collocation learning (Smadja & McKeown 1994) , all exploiting parallel corpora. To facilitate empirical studies that cannot rely on shared characteristics of Indo-European languages, we have been constructing the HKUST English-Chinese Parallel Bilingual Corpus.", |
|
"cite_spans": [ |
|
{ |
|
"start": 229, |
|
"end": 252, |
|
"text": "(Kay & R\u00f6scheisen 1988;", |
|
"ref_id": "BIBREF9" |
|
}, |
|
{ |
|
"start": 253, |
|
"end": 274, |
|
"text": "Catizone et al. 1989;", |
|
"ref_id": "BIBREF4" |
|
}, |
|
{ |
|
"start": 275, |
|
"end": 294, |
|
"text": "Gale & Church 1991;", |
|
"ref_id": "BIBREF8" |
|
}, |
|
{ |
|
"start": 295, |
|
"end": 313, |
|
"text": "Brown et al. 1991;", |
|
"ref_id": "BIBREF3" |
|
}, |
|
{ |
|
"start": 314, |
|
"end": 324, |
|
"text": "Chen 1993)", |
|
"ref_id": "BIBREF5" |
|
}, |
|
{ |
|
"start": 344, |
|
"end": 363, |
|
"text": "Fung & Church 1994)", |
|
"ref_id": "BIBREF7" |
|
}, |
|
{ |
|
"start": 398, |
|
"end": 417, |
|
"text": "(Brown et al. 1990;", |
|
"ref_id": "BIBREF1" |
|
}, |
|
{ |
|
"start": 418, |
|
"end": 436, |
|
"text": "Brown et al. 1993)", |
|
"ref_id": "BIBREF2" |
|
}, |
|
{ |
|
"start": 454, |
|
"end": 473, |
|
"text": "(Dagan et al. 1993)", |
|
"ref_id": "BIBREF6" |
|
}, |
|
{ |
|
"start": 502, |
|
"end": 520, |
|
"text": "(Gale et al. 1993)", |
|
"ref_id": "BIBREF8" |
|
}, |
|
{ |
|
"start": 548, |
|
"end": 571, |
|
"text": "(Smadja & McKeown 1994)", |
|
"ref_id": "BIBREF10" |
|
} |
|
], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "A Sentence-Aligned Corpus", |
|
"sec_num": "2" |
|
}, |
|
{ |
|
"text": "Currently, the material in the corpus consists primarily of fairly tight, literal sentence translations from the parliamentary proceedings of the Hong Kong Legislative Council. The original materials were not designed to be available in machine-readable form, and their obscure format necessitated heavy conversion and reformatting, using both manual and automatic processing. For the experiments reported in this paper, we began with a portion of the corpus occupying approximately 29Mb of raw English text and 15.5Mb of corresponding raw Chinese translation. The English text included nearly 5 million English words (Chinese words are hard to count, as discussed below).", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "A Sentence-Aligned Corpus", |
|
"sec_num": "2" |
|
}, |
|
{ |
|
"text": "The corpus is aligned using a hybrid statistical and lexical strategy. A dynamic programming algorithm optimizes an approximation to where \u0391 is an alignment, and T 1 and T 2 are the English and Chinese texts, respectively. An alignment A is a set consisting of L 1 \u2194 L 2 pairs where each L 1 or L 2 is an English or Chinese passage. The approximation depends largely on the lengths of the sentences, with some assistance from a very small set of high-reliability lexical cues:", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "A Sentence-Aligned Corpus", |
|
"sec_num": "2" |
|
}, |
|
{ |
|
"text": "where l 1 and l 2 are the lengths of the English and Chinese passages; v i is the number of occurrences of an English cue in L 1 and w i is the number of occurrences of the corresponding Chinese cue in L 2 and \u03b4 i (\u2022) is a normalized difference function that encapsulates the dependence within a single parameter for each i. For details, the reader is referred to Wu (1994) . The algorithm produces alignments with about 92% recall accuracy; an example is shown in Figure 1 .", |
|
"cite_spans": [ |
|
{ |
|
"start": 364, |
|
"end": 373, |
|
"text": "Wu (1994)", |
|
"ref_id": "BIBREF11" |
|
} |
|
], |
|
"ref_spans": [ |
|
{ |
|
"start": 465, |
|
"end": 473, |
|
"text": "Figure 1", |
|
"ref_id": "FIGREF0" |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "A Sentence-Aligned Corpus", |
|
"sec_num": "2" |
|
}, |
|
{ |
|
"text": "Some of the sentence alignments match different numbers of English and Chinese sentences, as in line 3 of Figure 1 . Such cases were discarded because they tend to be looser translations, and are thus harder to extract accurate correlations from. After retaining only 1-for-l sentence translations, the remaining data included approximately 17.9Mb of English (about 3 million words) and 9.6Mb of Chinese. The precision on the 1-for-l sentence pairs was approximately 96%, which turned out to be quite sufficient for the subsequent learning of translation lexicons.", |
|
"cite_spans": [], |
|
"ref_spans": [ |
|
{ |
|
"start": 106, |
|
"end": 114, |
|
"text": "Figure 1", |
|
"ref_id": "FIGREF0" |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "A Sentence-Aligned Corpus", |
|
"sec_num": "2" |
|
}, |
|
{ |
|
"text": "The Chinese portion of the corpus must be segmented before training, because written Chinese consists of a character stream with no space separators between words. Without segmentation, we would not know which Chinese character sequences are legitimate target chunks for translation. Segmentation is a somewhat arbitrary task though, since nearly all individual characters can be considered standalone words; the distinction between Chinese words, compounds, and collocations is unclear and may well be meaningless. In an attempt to circumvent this during an earlier phase of the project, we experimented with learning translation associations between English words and individual Chinese characters; while the results were encouraging, they were clearly unsatisfactory.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Training Procedure", |
|
"sec_num": "3" |
|
}, |
|
{ |
|
"text": "To segment the Chinese text, therefore, we used an online wordlist (BDC 1992) in conjunction with an optimization procedure described in Wu & Fung (1994) . Punctuation is separated out into word-level tokens as a byproduct of this process. According to the word segmentation produced by this method, the Chinese text consists of approximately 3.2 million words or tokens.", |
|
"cite_spans": [ |
|
{ |
|
"start": 137, |
|
"end": 153, |
|
"text": "Wu & Fung (1994)", |
|
"ref_id": "BIBREF12" |
|
} |
|
], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Training Procedure", |
|
"sec_num": "3" |
|
}, |
|
{ |
|
"text": "In addition, we wished to reduce noise from extraordinarily long sentences, which tend not to be translated sentence-by-sentence. We therefore removed all sentence pairs where either the English sentence was over 70 words long, or the Chinese sentence was over 90 words long. The English text was also normalized for punctuation, raising the word count to about 3.3 million tokens. After all prefiltering, the total input training text for the translation lexicon learning stage consisted of approximately 17.7Mb of English and 12.2Mb of Chinese.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Training Procedure", |
|
"sec_num": "3" |
|
}, |
|
{ |
|
"text": "The bilingual training process employs a variant of Brown et al.'s (1993) model, and as such is based on an iterative EM (expectation-maximization) procedure for maximizing the likelihood of generating the Chinese corpus given the English portion. The output of the training process is a set of potential Chinese translations for each English word, together with the probability estimate for each translation. The basic model of Brown et al. (1993) The training algorithm treats Equations 4-6 as re-estimation formulae for an iterative algorithm as follows:", |
|
"cite_spans": [ |
|
{ |
|
"start": 52, |
|
"end": 73, |
|
"text": "Brown et al.'s (1993)", |
|
"ref_id": null |
|
}, |
|
{ |
|
"start": 429, |
|
"end": 448, |
|
"text": "Brown et al. (1993)", |
|
"ref_id": "BIBREF2" |
|
} |
|
], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Training Procedure", |
|
"sec_num": "3" |
|
}, |
|
{ |
|
"text": "2. Compute the counts for all word translation pairs using Equation 4, summing over all sentence pairs in the corpus.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Choose any set of consistent initial values for t(\u2022).", |
|
"sec_num": "1." |
|
}, |
|
{ |
|
"text": "3. Compute the Lagrange multiplier for each English word using Equation 6.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Choose any set of consistent initial values for t(\u2022).", |
|
"sec_num": "1." |
|
}, |
|
{ |
|
"text": "4. Re-estimate the translation probabilities using Equation 5.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Choose any set of consistent initial values for t(\u2022).", |
|
"sec_num": "1." |
|
}, |
|
{ |
|
"text": "5. Repeat 2-4 until the translation probabilities converge.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Choose any set of consistent initial values for t(\u2022).", |
|
"sec_num": "1." |
|
}, |
|
{ |
|
"text": "For the corpus described here, training time is quite reasonable-approximately 24 hours on a Spare 10/51-to learn the Chinese translations for a total of 6,536 unique English words prior to filtering. A sample of the output is shown in Figure 2 . The most immediate practical use of the output is to have the lexicographer manually delete incorrect entries to produce a translation lexicon. Our first rough evaluation of learning performance is taken with respect to this application: it measures the percentage of English words for which a correct Chinese translation is found within the learned translation set. Of course this measure is meaningful only if the average size of the translation sets is fairly small. We therefore first prune the translation sets with the filters discussed in section 5, which eliminates many of the low-probability translations and thereby reduces the average size of the translation sets to 2.33 candidates per English word. Even after pruning, the resulting percentage correct is very high-95.1%-as estimated from a randomly drawn sample of 204 English words. 1 Encouraged by this result, we proceeded toward learning without manual correction. As a first pass, we evaluated the precision of the lexicon obtained by retaining only the single most probable translation for each English word. Another randomly drawn sample of 200 words yielded a precision estimate of 91.2% for this. This simple algorithm shows that even fully automatic procedures for English and Chinese are feasible with high precision.", |
|
"cite_spans": [ |
|
{ |
|
"start": 1096, |
|
"end": 1097, |
|
"text": "1", |
|
"ref_id": null |
|
} |
|
], |
|
"ref_spans": [ |
|
{ |
|
"start": 236, |
|
"end": 244, |
|
"text": "Figure 2", |
|
"ref_id": null |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "Baseline Performance", |
|
"sec_num": "4" |
|
}, |
|
{ |
|
"text": "Of course, most words have multiple potential translations, and the above method discards many correct alternative translations of English words. The problem is to find an automatic method for retaining these alternatives, without also retaining an excessive proportion of incorrect translations. and the translation probabilities are given by Figure 2 : Examples of unfiltered output with probabilities. Note that () is a special token lumping together all low-frequency Chinese words; Censorship is not correctly learned.", |
|
"cite_spans": [], |
|
"ref_spans": [ |
|
{ |
|
"start": 344, |
|
"end": 352, |
|
"text": "Figure 2", |
|
"ref_id": null |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "Baseline Performance", |
|
"sec_num": "4" |
|
}, |
|
{ |
|
"text": "The training procedure described above results in many translation entries with small or negligible probabilities, which should be pruned to produce a useful lexicon. An obvious solution to reduce noisy lexical entries is to set thresholds on probability. However, absolute thresholds work poorly. Sparse data causes many wrong entries to have inappropriately high probabilities. Conversely, some words are genuinely ambiguous and therefore legitimately spread out the probability across many translations. These cases should not be pruned by absolute thresholds.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Significance Filtering", |
|
"sec_num": "5" |
|
}, |
|
{ |
|
"text": "We therefore introduce two significance filtering criteria that simultaneously penalize for sparse data, and relax for ambiguous words. First, only English words that occur more than 25 times in the corpus are included in the lexicon. Second, for each word, only the translations accounting for the top 0.75 of the probability mass are retained; moreover, any translation with probability less than 0.11 is eliminated. In effect, the filtering threshold rises with data sparseness, and falls with the word's translation entropy.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Significance Filtering", |
|
"sec_num": "5" |
|
}, |
|
{ |
|
"text": "Evaluating the precision of this approach is slightly more involved, since alternative translation candidates have unequal probabilities. Note that after filtering, the probabilities of the candidates that remain are renormalized to sum to unity for each English word. We use these renormalized probabilities to weight the count of correct translations for the precision estimate. For example, if then we count this as 0.810 correct translations and 0.190 incorrect translations.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Significance Filtering", |
|
"sec_num": "5" |
|
}, |
|
{ |
|
"text": "Again, another random sample of 200 words was drawn, yielding a weighted precision estimate of 86.0%. Though less than the 91.2% figure for the earlier single-most-probable procedure, this precision is still quite high, even though each English word now has on average 2.33 Chinese translations. Some typical examples are shown in Figure 3 , where the highly-ranked translations of detect and Agreement are correct. Figure 3 also shows a range of the types of errors made. For words that occur frequently within frozen collocations, such as brain (as in brain drain) or empty, the probabilities for the correct translations can be too low. Her was incorrectly learned because the capitalized form is used predominantly in phrases such as Her Majesty's government, creating sometimes fail to be learned at all. Yet the large majority of entries followed the pattern exemplified by detect and Agreement.", |
|
"cite_spans": [], |
|
"ref_spans": [ |
|
{ |
|
"start": 331, |
|
"end": 339, |
|
"text": "Figure 3", |
|
"ref_id": "FIGREF2" |
|
}, |
|
{ |
|
"start": 416, |
|
"end": 424, |
|
"text": "Figure 3", |
|
"ref_id": "FIGREF2" |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "Significance Filtering", |
|
"sec_num": "5" |
|
}, |
|
{ |
|
"text": "Adaptiveness is one key strength of the statistical acquisition method; i.e., the translation learning is sensitive to the domain of translation. Translations for many governmental terms were picked up that would not normally be found in a hand-constructed translation lexicon. A more", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Significance Filtering", |
|
"sec_num": "5" |
|
}, |
|
{ |
|
"text": "We have described a series of techniques for automatically extracting a bilingual English-Chinese translation lexicon. The experiments reported here are, to our knowledge, the first large-scale empirical demonstrations of the applicability of pure statistical techniques to this task, and possibly the first between Indo-European and non-Indo-European languages. We have obtained high precision rates, between 86.0% and 95.1%, for lexicons of non-toy size, around 6,500 English words. Moreover, we have introduced a significance filtering method that can feed translation hypotheses of high precision into a manual post-filtering stage, or alternatively, yields excellent results even in fully automatic mode. A graph summarizing the precision results is shown in Figure 4 .", |
|
"cite_spans": [], |
|
"ref_spans": [ |
|
{ |
|
"start": 764, |
|
"end": 772, |
|
"text": "Figure 4", |
|
"ref_id": "FIGREF3" |
|
} |
|
], |
|
"eq_spans": [], |
|
"section": "Conclusion", |
|
"sec_num": "6" |
|
}, |
|
{ |
|
"text": "The statistical methods described are able to improve translation accuracy because the learned lexicons are finely adapted to the corpus domain. However, the methods still possess a strong non-adaptive component in the large hand-coded Chinese wordlist. Space limitations preclude description of an additional series of experiments that demonstrate the potential of adaptive techniques even more strongly. In the full paper (Wu & Xia 1994) , we show that even when the hand-coded wordlist is discarded, a fully adaptive procedure still raises performance significantly. ", |
|
"cite_spans": [ |
|
{ |
|
"start": 424, |
|
"end": 439, |
|
"text": "(Wu & Xia 1994)", |
|
"ref_id": "BIBREF13" |
|
} |
|
], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Conclusion", |
|
"sec_num": "6" |
|
}, |
|
{ |
|
"text": "Person names were excluded, but all other proper names were retained for this evaluation.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "", |
|
"sec_num": null |
|
} |
|
], |
|
"back_matter": [ |
|
{ |
|
"text": "We are indebted to Kenneth Church, William Gale, Pascale Fung, and the anonymous reviewers for helpful comments and suggestions, and to Eva Fong, Cindy Ng, and Derek Ngok for general contributions in the SILC project. The online Chinese wordlist (BDC 1992) was provided by Behavior Design Corporation.", |
|
"cite_spans": [], |
|
"ref_spans": [], |
|
"eq_spans": [], |
|
"section": "Acknowledgements", |
|
"sec_num": null |
|
} |
|
], |
|
"bib_entries": { |
|
"BIBREF0": { |
|
"ref_id": "b0", |
|
"title": "The BDC Chinese-English electronic dictionary (version 2.0). Behavior Design Corporation", |
|
"authors": [ |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Bdc", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1992, |
|
"venue": "", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "BDC. 1992. The BDC Chinese-English electronic dictionary (version 2.0). Behavior Design Corporation.", |
|
"links": null |
|
}, |
|
"BIBREF1": { |
|
"ref_id": "b1", |
|
"title": "A statistical approach to machine translation", |
|
"authors": [ |
|
{ |
|
"first": "Peter", |
|
"middle": [ |
|
"F" |
|
], |
|
"last": "Brown", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "John", |
|
"middle": [], |
|
"last": "Cocke", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Stephen", |
|
"middle": [ |
|
"A" |
|
], |
|
"last": "Dellapietra", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Vincent", |
|
"middle": [ |
|
"J" |
|
], |
|
"last": "Dellapietra", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Frederick", |
|
"middle": [], |
|
"last": "Jelinek", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "John", |
|
"middle": [ |
|
"D" |
|
], |
|
"last": "Lafferty", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Robert", |
|
"middle": [ |
|
"L" |
|
], |
|
"last": "Mercer", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "&", |
|
"middle": [], |
|
"last": "Paul S", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Roossin", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1990, |
|
"venue": "Computational Linguistics", |
|
"volume": "16", |
|
"issue": "2", |
|
"pages": "29--85", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "BROWN, PETER F., JOHN COCKE, STEPHEN A. DELLAPIETRA, VINCENT J. DELLAPIETRA, FREDERICK JELINEK, JOHN D. LAFFERTY, ROBERT L. MERCER, & PAUL S. ROOSSIN. 1990. A statistical approach to machine translation. Computational Linguistics, 16(2):29-85.", |
|
"links": null |
|
}, |
|
"BIBREF2": { |
|
"ref_id": "b2", |
|
"title": "The mathematics of statistical machine translation: Parameter estimation", |
|
"authors": [ |
|
{ |
|
"first": "Peter", |
|
"middle": [ |
|
"F" |
|
], |
|
"last": "Brown", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Stephen A", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Vincent", |
|
"middle": [ |
|
"J" |
|
], |
|
"last": "Dellapietra", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "&", |
|
"middle": [], |
|
"last": "Dellapietra", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Robert L", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Mercer", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1993, |
|
"venue": "Computational Linguistics", |
|
"volume": "19", |
|
"issue": "2", |
|
"pages": "263--311", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "BROWN, PETER F., STEPHEN A. DELLAPIETRA, VINCENT J. DELLAPIETRA, & ROBERT L. MERCER. 1993. The mathematics of statistical machine translation: Parameter estimation. Computational Lin- guistics, 19(2):263-311.", |
|
"links": null |
|
}, |
|
"BIBREF3": { |
|
"ref_id": "b3", |
|
"title": "Aligning sentences in parallel corpora", |
|
"authors": [ |
|
{ |
|
"first": "Peter", |
|
"middle": [ |
|
"F" |
|
], |
|
"last": "Brown", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Jennifer", |
|
"middle": [ |
|
"C" |
|
], |
|
"last": "Lai", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "&", |
|
"middle": [], |
|
"last": "Robert L", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Mercer", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1991, |
|
"venue": "Proceedings of the 29th Annual Conference of the Association for Computational Linguistics", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "169--176", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "BROWN, PETER F., JENNIFER C. LAI, & ROBERT L. MERCER. 1991. Aligning sentences in parallel corpora. In Proceedings of the 29th Annual Conference of the Association for Computational Linguistics, 169- 176, Berkeley.", |
|
"links": null |
|
}, |
|
"BIBREF4": { |
|
"ref_id": "b4", |
|
"title": "Deriving translation data from bilingual texts", |
|
"authors": [ |
|
{ |
|
"first": "Roberta", |
|
"middle": [], |
|
"last": "Catizone", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Graham", |
|
"middle": [], |
|
"last": "Russell", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "&", |
|
"middle": [], |
|
"last": "Susan Warwick", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1989, |
|
"venue": "Proceedings of the First International Acquisition Workshop", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "CATIZONE, ROBERTA, GRAHAM RUSSELL, & SUSAN WARWICK. 1989. Deriving translation data from bilingual texts. In Proceedings of the First International Acquisition Workshop, Detroit.", |
|
"links": null |
|
}, |
|
"BIBREF5": { |
|
"ref_id": "b5", |
|
"title": "Char_align: A program for aligning parallel texts at the character level", |
|
"authors": [ |
|
{ |
|
"first": "Stanley", |
|
"middle": [ |
|
"F" |
|
], |
|
"last": "Chen", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1993, |
|
"venue": "Proceedings of the 31st Annual Conference of the Association for Computational Linguistics", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "1--8", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "CHEN, STANLEY F. 1993. Aligning sentences in bilingual corpora using lexical information. In Proceedings of the 31st Annual Conference of the Association for Computational Linguistics, 9-16, Columbus, OH. CHURCH, KENNETH W. 1993. Char_align: A program for aligning parallel texts at the character level. In Proceedings of the 31st Annual Conference of the Association for Computational Linguistics, 1-8, Columbus, OH.", |
|
"links": null |
|
}, |
|
"BIBREF6": { |
|
"ref_id": "b6", |
|
"title": "Robust bilingual word alignment for machine aided translation", |
|
"authors": [ |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Dagan", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Kenneth", |
|
"middle": [ |
|
"W" |
|
], |
|
"last": "Ido", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "William", |
|
"middle": [ |
|
"A Gale" |
|
], |
|
"last": "Church", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1993, |
|
"venue": "Proceedings of the Workshop on Very Large Corpora", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "1--8", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "DAGAN, IDO, KENNETH W. CHURCH, & WILLIAM A. GALE. 1993. Robust bilingual word alignment for machine aided translation. In Proceedings of the Workshop on Very Large Corpora, 1-8, Columbus, OH.", |
|
"links": null |
|
}, |
|
"BIBREF7": { |
|
"ref_id": "b7", |
|
"title": "Statistical augmentation of a Chinese machine-readable dictionary", |
|
"authors": [ |
|
{ |
|
"first": "Pascale & Kenneth W", |
|
"middle": [], |
|
"last": "Fung", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Church", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1994, |
|
"venue": "Proceedings of the Fifteenth International Conference on Computational Linguistics", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "69--85", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "FUNG, PASCALE & KENNETH W. CHURCH. 1994. K-vec: A new approach for aligning parallel texts. In Proceedings of the Fifteenth International Conference on Computational Linguistics, 1096-1102, Kyoto. FUNG, PASCALE & DEKAI WU. 1994. Statistical augmentation of a Chinese machine-readable dictionary. In Proceedings of the Second Annual Workshop on Very Large Corpora, 69-85, Kyoto.", |
|
"links": null |
|
}, |
|
"BIBREF8": { |
|
"ref_id": "b8", |
|
"title": "A method for disambiguating word senses in a large corpus", |
|
"authors": [ |
|
{ |
|
"first": "William", |
|
"middle": [ |
|
"A" |
|
], |
|
"last": "Gale", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Kenneth W", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Church", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Berkeley", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "William", |
|
"middle": [ |
|
"A" |
|
], |
|
"last": "Gale", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "Kenneth", |
|
"middle": [ |
|
"W" |
|
], |
|
"last": "Church", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "&", |
|
"middle": [], |
|
"last": "David Yarowsky", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1991, |
|
"venue": "Proceedings of the 29th Annual Conference of the Association for Computational Linguistics", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "177--184", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "GALE, WILLIAM A. & KENNETH W. CHURCH. 1991. A program for aligning sentences in bilingual corpora. In Proceedings of the 29th Annual Conference of the Association for Computational Linguistics, 177- 184, Berkeley. GALE, WILLIAM A., KENNETH W. CHURCH, & DAVID YAROWSKY. 1993. A method for disambiguating word senses in a large corpus. In Computers and the Humanities.", |
|
"links": null |
|
}, |
|
"BIBREF9": { |
|
"ref_id": "b9", |
|
"title": "Text-translation alignment", |
|
"authors": [ |
|
{ |
|
"first": "Martin & M", |
|
"middle": [], |
|
"last": "Kay", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "R\u00f6scheisen", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1988, |
|
"venue": "", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "KAY, MARTIN & M. R\u00d6SCHEISEN. 1988. Text-translation alignment. Technical Report P90-00143, Xerox Palo Alto Research Center.", |
|
"links": null |
|
}, |
|
"BIBREF10": { |
|
"ref_id": "b10", |
|
"title": "Translating collocations for use in bilingual lexicons", |
|
"authors": [ |
|
{ |
|
"first": "Frank", |
|
"middle": [ |
|
"A" |
|
], |
|
"last": "Smadja", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Kathleen R", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Mckeown", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1994, |
|
"venue": "Proceedings of the ARPA Human Language Technology Workshop", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "SMADJA, FRANK A. & KATHLEEN R. MCKEOWN. 1994. Translating collocations for use in bilingual lexicons. In Proceedings of the ARPA Human Language Technology Workshop, Princeton, N.J.", |
|
"links": null |
|
}, |
|
"BIBREF11": { |
|
"ref_id": "b11", |
|
"title": "Aligning a parallel English-Chinese corpus statistically with lexical criteria", |
|
"authors": [ |
|
{ |
|
"first": "Dekai", |
|
"middle": [], |
|
"last": "Wu", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1994, |
|
"venue": "Proceedings of the 32nd Annual Conference of the Association for Computational Linguistics", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "80--87", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "WU, DEKAI. 1994. Aligning a parallel English-Chinese corpus statistically with lexical criteria. In Proceedings of the 32nd Annual Conference of the Association for Computational Linguistics, 80-87, Las Cruces, New Mexico.", |
|
"links": null |
|
}, |
|
"BIBREF12": { |
|
"ref_id": "b12", |
|
"title": "Improving Chinese tokenization with linguistic filters on statistical lexical acquisition", |
|
"authors": [ |
|
{ |
|
"first": "Dekai & Pascale", |
|
"middle": [], |
|
"last": "Wu", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Fung", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1994, |
|
"venue": "Proceedings of the Fourth Conference on Applied Natural Language Processing", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "WU, DEKAI & PASCALE FUNG. 1994. Improving Chinese tokenization with linguistic filters on statistical lexical acquisition. In Proceedings of the Fourth Conference on Applied Natural Language Processing, Stuttgart. To appear.", |
|
"links": null |
|
}, |
|
"BIBREF13": { |
|
"ref_id": "b13", |
|
"title": "Large-scale automatic extraction of an English-Chinese lexicon", |
|
"authors": [ |
|
{ |
|
"first": "Dekai & Xuanyin", |
|
"middle": [], |
|
"last": "Wu", |
|
"suffix": "" |
|
}, |
|
{ |
|
"first": "", |
|
"middle": [], |
|
"last": "Xia", |
|
"suffix": "" |
|
} |
|
], |
|
"year": 1994, |
|
"venue": "", |
|
"volume": "", |
|
"issue": "", |
|
"pages": "", |
|
"other_ids": {}, |
|
"num": null, |
|
"urls": [], |
|
"raw_text": "WU, DEKAI & XUANYIN XIA, 1994. Large-scale automatic extraction of an English-Chinese lexicon. Sub- mitted.", |
|
"links": null |
|
} |
|
}, |
|
"ref_entries": { |
|
"FIGREF0": { |
|
"text": "A sample of length-based alignment output.", |
|
"type_str": "figure", |
|
"num": null, |
|
"uris": null |
|
}, |
|
"FIGREF1": { |
|
"text": "assumes that the probability of translating a given English sentence e = e 1 e 2 .\u2022\u2022\u2022e l into a Chinese sentence c = c 1 c 2 \u2022\u2022\u2022 c m following a particular word alignment a = a 1 a 2 \u2022\u2022\u2022a m can be approximated by where t(\u2022) are translation probabilities for individual word pairs and \u03b5 is a small constant. Under this assumption, the expected number of times that any particular word e in an English training sentence e generates any particular word c in the corresponding Chinese training sentence c is given by is the Lagrange multiplier for word e.", |
|
"type_str": "figure", |
|
"num": null, |
|
"uris": null |
|
}, |
|
"FIGREF2": { |
|
"text": "Examples of significance-filtered output with probabilities.", |
|
"type_str": "figure", |
|
"num": null, |
|
"uris": null |
|
}, |
|
"FIGREF3": { |
|
"text": "Summary of precision results.", |
|
"type_str": "figure", |
|
"num": null, |
|
"uris": null |
|
} |
|
} |
|
} |
|
} |