Improving Domain-Specific Word Alignment with a General Bilingual
Contents
1. Alignment Models In Proc of the 38th Annual Meeting of the Association for Computational Linguistics ACL 2000 440 447 11 Smadja F McKeown K R Hatzivassiloglou V Translating Collocations for Bilingual Lexicons a Statistical Approach Computational Linguistics 1996 Vol 22 No 1 1 38 12 Simard M Langlais P Sub sentential Exploitation of Translation Memories In Proc of MT Summit VIII 2001 335 339 13 Somers H Review Article Example Based Machine Translation Machine Translation 1999 Vol 14 No 2 113 157 14 Tufis D Barbu A M Lexical Token Alignment Experiments Results and Application In Proc of the Third Int Conf on Language Resources and Evaluation LREC 2002 458 465 15 Wu D K Stochastic Inversion Transduction Grammars and Bilingual Parsing of Parallel Corpora Computational Linguistics 1997 Vol 23 No 3 377 4032. Thus the subtraction set contains two different alignment links for each English word For the specific domain we use SF and SF to represent the word alignment sets in the two directions The symbols SF PF and MF represents the intersection set union set and the subtraction set respectively 3 2 Translation Dictionary Acquisition When we train the statistical word alignment model with the large scale bilingual corpus in the general domain we can get two word alignment results for the training data By taking the intersection of the two word alignment results we build a new alignment set The alignment links in this intersection set are extended by iteratively adding word alignment links into it as described in 10 Based on the extended alignment links we build an English to Chinese translation dictionary D with translation probabilities In order to filter some noise caused by the error alignment links we only retain those translation pairs whose translation probabilities are above a threshold 6 or co occurring frequencies are above a threshold When we train the IBM statistical word alignment model with the small scale bilingual corpus in the specific domain we build another translation dictionary D with the same method as for the dictionary D But we adopt a different filtering strategy for the translation dictionary D We use log likelihood ratio to estimate the association strength of each translation pair bec
3. Improving Domain Specific Word Alignment with a General Bilingual Corpus WU Hua WANG Haifeng Toshiba China Research and Development Center 5 F Tower W2 Oriental Plaza No 1 East Chang An Ave Dong Cheng District Beijing 100738 China wuhua wanghaifeng rdc toshiba com cn Abstract In conventional word alignment methods some employ statistical models or statistical measures which need large scale bilingual sentence aligned training corpora Others employ dictionaries to guide alignment selection However these methods achieve unsatisfactory alignment results when performing word alignment on a small scale domain specific bilingual corpus without terminological lexicons This paper proposes an approach to improve word alignment in a specific domain in which only a small scale domain specific corpus is available by adapting the word alignment information in the general domain to the specific domain This approach first trains two statistical word alignment models with the large scale corpus in the general domain and the small scale corpus in the specific domain respectively and then improves the domain specific word alignment with these two models Experimental results show a significant improvement in terms of both alignment precision and recall achieving a relative error rate reduction of 21 96 as compared with state of the art technologies 1 Introduction Bilingual word alignment is first introduced as an intermediate resul
4. ause Dunning 5 proved that log likelihood ratio performed very well on small scale data Thus we get the translation dictionary D by keeping those entries whose log likelihood ratio scores are greater than a threshold 63 The corpus used to build D is the 320 000 sentence pairs in the general domain The corpus used to build D is the 366 sentence pairs on the manual for a medical system By setting thresholds 0 1 5 5 and 6 50 we get two translation dictionaries the statistics information of which is showed in Table 2 4 Table 2 Translation Dictionary Statistics D D Unique English Words 57 380 728 Multi Words 18 870 28 Average Chinese Translations 2 1 1 1 4 The thresholds are obtained to ensure the best compromise of alignment precision and recall on the testing set In the translation dictionary D the multi words accounts for 32 89 of the total words In the translation dictionary D the number of multi words is small because the training data are very limited 3 3 Word Alignment Improvement With the statistical word alignment models and the translation dictionaries trained on the corpora in the general domain and the specific domain we describe the algorithm to improve the domain specific word alignment in this section Based on the bi directional word alignment we define SJ as SI SG SF and UG as UG PG U PF 4 SI The word alignment links in the set SZ are very reliable Thus we directly accept
5. cale general bilingual corpus while the available domain specific bilingual corpus is usually quite small Thus we use the bilingual corpus in the general domain to improve word alignments for general words and the bilingual corpus in the specific domain for domain specific words In other words we will adapt the word alignment information in the general domain to the specific domain Although the adaptation technology is widely used for other tasks such as language modeling few literatures to the best of our knowledge directly address word alignment adaptation The work most closely related to ours is the statistical translation adaptation described in 7 Langlais used terminological lexicons to improve the performance of a statistical translation engine which is trained on a general bilingual corpus and used to translate a manual for military snipers The experimental results showed that this adaptation method could reduce word error rate on the translation task In this paper we perform word alignment adaptation from the general domain to a specific domain in this study a user manual for a medical system with four steps 1 We train a word alignment model using a bilingual corpus in the general domain 2 We train another word alignment model using a small scale bilingual corpus in the specific domain 3 We build two translation dictionaries according to the alignment results in 1 and 2 respectively 4 For each sentence pair in th
6. ce system scanning volume Pa t Pa z ig 7 t Ie es z P 7 n 1 ee r Pee EE ET Fig 1 Alignment Example Based on the above analysis it can be seen that it is not effective to directly combine the bilingual corpus in the general domain and in the specific domain as training data However the correct alignment links extracted by the method G and those extracted by the method S are complementary to each other Thus we can develop a method to improve the domain specific word alignment based on the results of both the method G and the method S Another kind of errors is about the multi word alignment links The IBM statistical word alignment model only allows one to one or more to one alignment links However the domain specific terms are usually aligned to more than one Chinese word Thus the multi word unit in the corpus cannot be correctly aligned using this statistical model For this case we will use translation dictionaries as guides to modify some alignment links and get multi word alignments 3 Word Alignment Adaptation According to the result analysis in Section 2 3 we take two measures to improve the word alignment results One is to combine the word alignment results of both the method G and the method S The other is to use translation dictionaries 3 1 Bi directional Word Alignment In statistical translation models 3 only one to one and more to one word alignment li
7. corpus in the general domain which includes 320 000 bilingual sentence pairs and a sentence aligned English Chinese bilingual corpus in the specific domain a user manual for a medical system which includes 546 bilingual sentence pairs From this domain specific corpus we randomly select 180 pairs as testing data The remained 366 pairs are used as domain specific training data The Chinese sentences in both the training set and the testing set are automatically segmented into words Thus there are two kinds of errors for word alignment one is the word segmentation error and the other is the alignment error In Chinese if a word is incorrectly segmented the alignment result is also incorrect For example for the Chinese sentence i2 Dit PR H HE ft x2 Warning label for the couch top our system segments it into i r R H F E E The sequence PRIM HAY is incorrectly segmented into P M M couch taxi which should be R fil f J couch top of Thus the segmentation errors in Chinese may change the word meaning which in turn cause alignment errors In order to exclude the effect of the segmentation errors on our alignment results we correct the segmentation errors in our testing set The alignments in the testing set are manually annotated which includes 1 478 alignment links 2 2 Overall Performance There are several different evaluation methods f
8. e specific domain we use the two models to get different word alignment results and improve the results according to the translation dictionaries Experimental results show that our approach improves domain specific word alignment in terms of both precision and recall achieving a 21 96 relative error rate reduction The remainder of the paper is organized as follows Section 2 introduces the statistical word alignment method and analyzes the problems existing in this method for the domain specific task Section 3 describes our word alignment adaptation algorithm Section 4 describes the evaluation results The last section concludes our approach and presents the future work 2 Statistical Word Alignment In this section we apply the IBM statistical word alignment models to our domain specific corpus and analyze the alignment results The tool used for statistical word alignment is GIZA 10 With this tool we compare the word alignment results of three methods These methods use different corpora to train IBM word alignment model 4 The method G S directly combines the bilingual sentence pairs in the general domain and in the specific domain as training data The method G only uses the bilingual sentence pairs in the general domain as training data The method S only uses the bilingual sentence pairs in the specific domain as training data 2 1 Training and Testing Data We have a sentence aligned English Chinese bilingual
9. en only uses the corpus in the general domain as training data The third method Spec only uses the domain specific corpus as training data With these training data the three methods can get their own translation dictionaries However each of them can only get one translation dictionary Thus only one of the two steps a and b in Figure 2 can be applied to these methods All of these three methods first get bi directional statistical word alignment using the GIZA tool and then use the trained translation dictionary to improve the statistical word alignment results The difference between these three methods and our method is that for each source word our method provides four candidate alignment links while the other three methods only provides two candidate alignment links Thus the steps c and d in Figure 2 cannot be applied to these three methods The training data and the testing data are the same as described in Section 2 1 With the evaluation metrics described in section 2 2 we get the alignment results shown in Table 3 From the results it can be seen that our approach performs the best among others Our method achieves a 21 96 relative error rate reduction as compared with the method Gen Spec In addition by comparing the results in Table 3 and those in Table 1 in Section 2 2 we can see that the precision of word alignment links is improved by using the translation dictionaries Thus introducing translation dictio
10. enberg L Merkel M Hein A S Tiedemann J Evaluation of Word Alignment Systems In Proc of the Second Int Conf on Linguistic Resources and Evaluation LREC 2000 1255 1261 3 Brown P F Della Pietra S Della Pietra V Mercer R The Mathematics of Statistical Machine Translation Parameter estimation Computational Linguistics 1993 Vol 19 No 2 263 311 4 Cherry C Lin D K A Probability Model to Improve Word Alignment In Proc of the 41st Annual Meeting of the Association for Computational Linguistics ACL 2003 88 95 5 Dunning T Accurate Methods for the Statistics of Surprise and Coincidence Computational Linguistics 1993 Vol 19 No 1 61 74 6 Ker S J Chang J S A Class based Approach to Word Alignment Computational Linguistics 1997 Vol 23 No 2 313 343 7 Langlais P Improving a General Purpose Statistical Translation Engine by Terminological Lexicons In Proc of the 2nd Int Workshop on Computational Terminology COMPUTERM 2002 1 7 8 Melamed D Automatic Construction of Clean Broad Coverage Translation Lexicons In Proc of the 2nd Conf of the Association for Machine Translation in the Americas AMTA 1996 125 134 9 Menezes A Richardson S D A Best First Alignment Algorithm for Automatic Extraction of Transfer Mappings from Bilingual Corpora In Proc of the ACL 2001 Workshop on Data Driven Methods in Machine Translation 2001 39 46 10 Och F J Ney H Improved Statistical
11. ion Recall AER G S 0 7140 0 6942 0 2961 G 0 7136 0 6847 0 3014 S 0 4486 0 4066 0 5735 2 3 Result Analysis We use A B and C to represent the set of correct alignment links extracted by the method G S the method G and the method S respectively From the experiments we get 4 1026 B 1012 and C 601 and get two intersection sets D ANC 524 and E BOC 516 Thus about 14 alignment links of C are not covered by B That is to say although the size of the domain specific corpus is very small it can produce word alignment links that are not covered by the general corpus These alignment links usually include domain specific words Moreover about 13 alignment links of C are not covered by A This indicates that by combining the two corpora the method G S still cannot detect the domain specific alignment links At the same time about 49 of alignment links in both A and B are not covered by the set C For example in the sentence pair in Figure 1 there is a domain specific word multislice For this word both the method G S and G produce a wrong alignment link multislice 414i while the method S produces a correct word alignment link multislice 4 414i However the general word alignment link refer to amp JL is detected by both the method G S and the method G but not detected by the method S Refer to the multisli
12. nary results in alignment precision improving while combining the alignment results of Gen and Spec results in alignment recall improving Table 3 Word Alignment Adaptation Results Method Precision Recall AER Ours 0 8363 0 7673 0 1997 Gen Spec 0 8276 0 6758 0 2559 Gen 0 8668 0 6428 0 2618 Spec 0 8178 0 4769 0 3974 Table 4 Multi Word Alignment Results Method Precision Recall AER Ours 0 5665 0 4083 0 5254 Gen Spec 0 4339 0 096 0 8430 Gen 0 5882 0 083 0 8541 Spec 0 5854 0 100 0 8292 In the testing set there are 240 multi word alignment links Most of the links consist of domain specific words Table 4 shows the results for multi word alignment Our method achieves much higher recall than the other three methods and achieves comparable precision This indicates that combining the alignment results created by the Gen method and the Spec method increases the possibility of obtaining multi word alignment links From the table it can be also seen that the Spec method performs better than both the Gen method and the Gen Spec method on the multi word alignment This indicates that the Spec method can catch domain specific alignment links even when trained on the small scale corpus It also indicates that by adding the domain specific data into the general training data the method Gen Spec cannot catch the domain specific alignment links 5 Conclusion and Future Work This paper pro
13. nks can be found Thus some multi word units cannot be correctly aligned In order to deal with this problem we perform translation in two directions English to Chinese and Chinese to English as described in 10 The GIZA toolkit is used to perform statistical word alignment For the general domain we use SG and SG to represent the alignment sets obtained with English as the source language and Chinese as the target language or vice versa For alignment links in both sets we use i for English words and j for Chinese words SG 4 4 a a 2 0 5 SG3 i 4 4 aj a 20 6 Where a x i j represents the index position of the source word aligned to the target word in position x For example if a Chinese word in position j is connected to an English word in position i then a i Ifa Chinese word in position j is connected to English words in positions i and i then 4 i i2 Based on the two alignment sets we obtain their intersection set union set and subtraction set gt Multi word alignment links means one or more source words aligned to more than one target word or vice versa In this paper the union operation does not remove the replicated elements For example if set one includes two elements 1 2 and set two includes two elements 1 3 then the union of these two sets becomes 1 1 2 3 Intersection SG SG ASG Union PG SG USG Subtraction MG PG 2 SG
14. or word alignment 2 In our evaluation we use evaluation metrics similar to those in 10 However we do not classify alignment links into sure links and possible links We consider each alignment s f as a sure link where both s and can be words or multi word units If we use S to represent the alignments identified by the proposed methods and Sc to denote the reference alignments the methods to calculate the precision recall and f measure are shown in Equation 1 2 and 3 According to the definition of the alignment error rate AER in 10 AER can be calculated with Equation 4 Thus the higher the f measure is the lower the alignment error rate is ISg Sc precision 1 ISo recall 86 0501 2 Sc fineasure 2 Se OSe 3 ISgl Sc AER 1 2S SGN 1 fmeasure 4 ale ise Generally a user manual only includes several hundred sentences With the above metrics we evaluate the three methods on the testing set with Chinese as the source language and English as the target language The results are shown in Table 1 It can be seen that although the method G S achieves the best results among others it performs just a little better than the method G This indicates that adding the small scale domain specific training sentence pairs into the general corpus doesn t greatly improve the alignment performance Table 1 Statistical Word Alignment Results Method Precis
15. poses an approach to improve domain specific word alignment through alignment adaptation Our contribution is that given a large scale general bilingual corpus and a small scale domain specific corpus our approach improves the domain specific word alignment results in terms of both precision and recall In addition with the training data two translation dictionaries are built to select or modify the word alignment links and to further improve the alignment results Experimental results indicate that our approach achieves a precision of 83 63 and a recall of 76 73 for word alignment on the manual of a medical system resulting in a relative error rate reduction of 21 96 This indicates that our method significantly outperforms the method only combining the general bilingual corpus and the domain specific bilingual corpus as training data Our future work includes two aspects First we will seek other adaptation methods to further improve the domain specific word alignment results Second we will also use the alignment results to build terminological lexicons and to improve translation quality and efficiency in machine translation systems References 1 Ahrenberg L Merkel M Andersson M A Simple Hybrid Aligner for Generating Lexical Correspondences in Parallel Tests In Proc of the 36th Annual Meeting of the Association for Computational Linguistics and the 17th Int Conf on Computational Linguistics ACL COLING 1998 29 35 2 Ahr
16. t in statistical machine translation SMT 3 Besides being used in SMT it is also used in translation lexicon building 8 transfer rule learning 9 example based machine translation 13 translation memory systems 12 etc In previous alignment methods some researchers modeled the alignments as hidden parameters in a statistical translation model 3 10 or directly modeled them given the sentence pairs 4 Some researchers use similarity and association measures to build alignment links 1 11 14 In addition Wu 15 used a stochastic inversion transduction grammar to simultaneously parse the sentence pairs to get the word or phrase alignments However All of these methods require a large scale bilingual corpus for training When the large scale bilingual corpus is not available some researchers use existing dictionaries to improve word alignment 6 However few works address the problem of domain specific word alignment when neither the large scale domain specific bilingual corpus nor the domain specific translation dictionary is available In this paper we address the problem of word alignment in a specific domain in which only a small scale corpus is available In the domain specific corpus there are two kinds of words Some are general words which are also frequently used in the general domain Others are domain specific words which only occur in the specific domain In general it is not quite hard to obtain a large s
17. them as correct links and add them into the final alignment set WA In the set UG there are two to four different alignment links for each word We first examine the dictionary D and then D to see whether there is at least one alignment link of this word included in these two dictionaries If it is successful we add the link with the largest probability or the largest log likelihood ratio score to the final set WA Otherwise we use two heuristic rules to select alignment links The detailed algorithm is described in Figure 2 Input Alignment sets S7 and UG 1 For alignment links in SZ we directly add them into WA 2 For each English word i we first find its alignment links in UG and then do the following a If there are alignment links found in the translation dictionary D we add the link with the largest probability to WA b Otherwise if there are alignment links found in the translation dictionary D we add the link with the largest log likelihood ratio score to WA If both a and b fail but three links select the same target words for the English word i we add this link to WA d Otherwise if there are two different kinds of links for this word one target is a single word and the other target is a multi word unit and the words in the multi word unit have no link in WA add this multi word alignment link to WA Output Updated alignment set WA c sf Fig 2 Word Alignment Adaptation Algori
18. thm Figure 3 lists four examples for word alignment adaptation In example 1 the phrase based on has two different alignment links one is based on 4 amp F and the other is based 4 F And in the translation dictionary D the phrase based on can be translated into J Thus the link based on 4 F is finally selected according to rule a in Figure 2 In the same way the link contrast i i3 in example 2 is selected with the translation dictionary D The link reconstructed 4 3 in Example 3 is obtained because there are three alignment links selecting it For the English word x ray in Example 4 we have two different links in UG One is x ray X and the other is x ray X JZ And the single Chinese words and 2 have no alignment links in the set WA According to the rule d we select the link x ray X H2 1 Linkage i is made based on the related information HER ssf ae we a RER AT Pa 2 The CT number depends on contrast enhancement level fo rA 7 Pea zo aa 7 a CT RE BEE aw wa eE z 3 The econsk cmyi image cannot be obtained Fig 3 Alignment Adaptation Example 4 Evaluation In this section we compare our methods with three other methods The first method Gen Spec directly combines the corpus in the general domain and in the specific domain as training data The second method G
Download Pdf Manuals
Related Search