CN107357783A - A kind of English translation mass analysis method of translator of Chinese into English - Google Patents

Abstract

The invention provides a method for analyzing the quality of an English translation from Chinese to English. The method is a translation English translation preprocessing module, a translation English translation fidelity analysis module, a translation English translation semantic similarity analysis module, and a translation English translation. An analysis model composed of translation quality analysis result generation modules. After an English translation is processed by the analysis model, the quality analysis results of the English translation can be obtained at last. The method of the invention solves the problem of automatically analyzing the quality of the English translation translated from Chinese into English, and its analysis result is better than that of the traditional method for analyzing the quality of the English translation translated from Chinese into English.

Description

A method for quality analysis of English translations from Chinese to English

technical field

The invention relates to statistical natural language processing technology, word grammar analysis technology in translation from Chinese to English, content analysis technology in translation from Chinese to English, and specifically a quality analysis method for English translation from Chinese to English.

Background technique

Traditional Chinese-to-English English translation quality analysis methods mainly include exact matching method, edit distance method, and N-gram-based method. The exact match method has too strict requirements on the English translation from Chinese to English. Only when the content of the translated English translation is exactly the same as the content of the standard English translation, the translated English translation is judged to be correct. This method completely ignores a part of the English word. word polymorphism problem. The edit distance method is to analyze the quality of the English translation from Chinese to English: check the number of insertions, deletions, and replacements of English words in the process of converting the English translation into a standard English translation, and then analyze the translation quality of the English translation based on these times. However, this method cannot check whether words between the translated English translation and the standard English translation are synonyms. The N-gram method is to match the N-tuples that appear in the translated English translation with the N-tuples that appear in the standard English translation, and analyze the translation quality of the English translation according to the number of successful matches. However, this method cannot check the translation The semantic similarity between the English translation and the standard English translation. In order to solve the above problems, the present invention provides a method for analyzing the quality of English translations from Chinese to English.

Contents of the invention

The overall processing flow of the English translation quality analysis method of Chinese translation into English of the present invention is as shown in Figure 1, wherein includes translation English translation preprocessing module, translation English translation fidelity analysis module, translation English translation semantic similarity analysis module, translation English translation quality analysis result generation module.

The processing flow of the translated English translation preprocessing module is as follows: first, input the translated English translation and the standard English translation, respectively perform word segmentation, lowercase words, and remove stop words on the translated English translation and the standard English translation; Part-of-speech tagging, part-of-speech disambiguation, and phrase segmentation of the translated English translation and standard English translation of word segmentation, word lowercase, and stop word removal; third, output the preprocessing results of the translated English translation and standard English translation of the above two-step processing .

The processing flow of the translation English translation loyalty analysis module is as follows: first, input translation English translation, standard English translation, translation English translation preprocessing result, standard English translation preprocessing result, number translation English translation, standard English translation, translation Words in the preprocessing results of the English translation and the preprocessing results of the standard English translation; second, perform an exact match between the words in the translated English translation and the words in the standard English translation, and record the words and their numbers that match successfully; The translated English translation and the standard English translation exactly match the remaining words for stem matching and record the words and numbers that match successfully; fourth, perform synonym matching for the remaining words between the translated English translation and the standard English translation for stem matching and record the matching success Fifth, use the word matching rate accuracy weights generated by the training translation English translation set and the standard English translation set and the total number of successfully matched words in the translated English translation to calculate and output the fidelity of the translated English translation.

The processing flow of the translation English translation semantic similarity analysis module is as follows: first, input the translated English translation and standard English translation part-of-speech disambiguation results; second, search and record the corresponding words in the English semantic dictionary according to the part-of-speech disambiguation results Node number; third, in the English semantic dictionary, find the node number corresponding to the standard English word from the node number corresponding to the translated English word, and record the node number corresponding to each word on the search path and the directed edges between the nodes Fourth, according to the node number corresponding to each word on the search path and the probability of directed edges between nodes, generate a two-dimensional vector of word probability distribution in the translated English translation, and iteratively calculate the word probability distribution in the standard English translation Two-dimensional vector; Fifth, merge the two-dimensional vector of word probability distribution in the translated English translation to obtain the two-dimensional vector of the probability distribution of the translated English translation, and extract the word probability value from the two-dimensional vector of the probability distribution of the translated English translation to generate the probability distribution of the translated English translation One-dimensional vector; sixth, combine the standard English translation word probability distribution two-dimensional vector to obtain the standard English translation probability distribution two-dimensional vector, and extract the word probability value from the standard English translation probability distribution two-dimensional vector to generate the standard English translation probability distribution one Seventh, calculate and output the semantic similarity of the translated English translation according to the one-dimensional vector of the probability distribution of the translated English translation and the one-dimensional vector of the probability distribution of the standard English translation.

The processing flow of the translation English translation quality analysis result generation module is as follows: first, input the output results of the translation English translation fidelity analysis module, and the output results of the translation English translation semantic similarity analysis module; second, according to the translation English translation faithfulness The output result of the degree analysis module and the output result of the semantic similarity analysis module of the translated English translation are used to generate the quality score and comments of the translated English translation.

The definition of the invention is as follows:

1. Word part-of-speech tagging set

The present invention adopts the Penn State Treebank tagging set of the University of Pennsylvania in the United States, and carries out word part-of-speech tagging on the translated English translation and the standard English translation according to the rules of the tagging set.

2. Word part-of-speech tagging structure

Word part-of-speech tagging is the part-of-speech tagging process for words in the translated English translation and the standard English translation. The following is the format after the part-of-speech tagging:

Word ₁ [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...] Word ₂ [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...]...

Word _n [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...]

3. Phrase block structure

Phrase segmentation is to segment the noun phrases and verb phrases in the translated English translation and the standard English translation. The following is the format of the phrase segmentation:

word ₁ /phrase cut ₁ , word ₂ /phrase cut ₂ , ... word _n /phrase cut _n

4. Part-of-speech disambiguation structure

Part-of-speech disambiguation refers to comparing the part-of-speech tagging results of the translated English translation and the standard English translation with the English vocabulary online dictionary, and removing the incorrectly tagged word part-of-speech tagging results. The following is the format after the part-of-speech disambiguation:

Word ₁ [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...] Word ₂ [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...]...

Word _n [Part of Speech ₁ #Part of Speech ₂ #Part of Speech ₃ ...]

5. English Semantic Dictionary

The English semantic dictionary refers to a semantic dictionary that organizes entry information in alphabetical order and is based on multiple interpretations of words. The dictionary includes five parts of common nouns, verbs, adjectives, adverbs, and function words. The format of the English semantic dictionary is as follows:

Word ₁ [part of speech ₁ ] [word frequency ₁ ] [offset ₁ ] [dictionary file detailed path ₁ ] [word meaning interpretation ₁ ] [word meaning corresponding node number ₁ ] [directed edge probability between nodes ₁ ]

Word ₁ [part of speech ₂ ] [word frequency ₂ ] [offset ₂ ] [dictionary file detailed path ₂ ] [word meaning interpretation ₂ ] [word meaning corresponding node number ₂ ] [directed edge probability between nodes ₂ ]

...

Word ₁ [part of speech _n ][word frequency _n ][offset _n ][detailed path of dictionary file _n ][word meaning and definition _n ][word meaning corresponding node number _n ][directed edge probability _n between nodes]

Word ₂ [part of speech ₁ ] [word frequency ₁ ] [offset ₁ ] [dictionary file detailed path ₁ ] [word meaning interpretation ₁ ] [word meaning corresponding node number ₁ ] [directed edge probability between nodes ₁ ]

Word ₂ [part of speech ₂ ] [word frequency ₂ ] [offset ₂ ] [dictionary file detailed path ₂ ] [word meaning interpretation ₂ ] [word meaning corresponding node number ₂ ] [directed edge probability between nodes ₂ ]

...

Word ₂ [part of speech _n ][word frequency _n ][offset _n ][detailed path of dictionary file _n ][word meaning definition _n ][word meaning corresponding node number _n ][directed edge probability _n between nodes]

Word _n [part of speech ₁ ] [word frequency ₁ ] [offset ₁ ] [dictionary file detailed path ₁ ] [word meaning interpretation ₁ ] [word meaning corresponding node number ₁ ] [directed edge probability between nodes ₁ ]

Word _n [part of speech ₂ ] [word frequency ₂ ] [offset ₂ ] [dictionary file detailed path ₂ ] [word meaning interpretation ₂ ] [word meaning corresponding node number ₂ ] [directed edge probability between nodes ₂ ]

...

Word _n [part of speech _n ] [word frequency _n ] [offset _n ] [dictionary file detailed path _n ] [word meaning definition _n ] [word meaning corresponding node number _n ] [directed edge probability _n between nodes]

6. Word matching accuracy calculation formula

Word matching accuracy refers to the ratio of the number of words successfully matched between the translated English translation and the standard English translation to the total number of words in the translated English translation. Its calculation formula is as follows:

7. Word matching recall calculation formula

Word matching recall refers to the ratio of the number of successful word matches between the translated English translation and the standard English translation to the total number of words in the standard English translation. Its calculation formula is as follows:

8. Calculation formula of harmonic mean

The harmonic mean refers to the average value between the word matching accuracy rate of calculation formula (1) and the word matching recall rate of calculation formula (2), and its calculation formula is as follows:

In the calculation formula (3), the word matching success rate is calculated by the calculation formula (1), the word matching recall rate is calculated by the calculation formula (2), a is the weight of the word matching accuracy rate, and 1-a is the word Weight for matching recall.

9. Calculation formula of penalty coefficient value

The penalty coefficient value is a coefficient to reduce the error between the content of the translated English translation and the content of the standard English translation caused by the harmonic mean, and its calculation formula is as follows:

In the calculation formula (4), the number of phrases refers to the total number of noun phrases and verb phrases in the translated English translation; b is the adjustment parameter of the penalty coefficient value; c is the ratio of the number of phrases to the number of words that are successfully matched in the translated English translation The adjustment parameters of b and c are both calculated by the English translation fidelity analysis module.

10. Calculation formula for English translation fidelity

The English translation fidelity refers to the semantic similarity between the words in the translated English translation and the words in the standard English translation. Its calculation formula is as follows:

English translation fidelity = (1- penalty coefficient value) × harmonic mean (5)

In the calculation formula (5), the harmonic mean value is calculated by the calculation formula (3), and the penalty coefficient value is calculated by the calculation formula (4).

11. Probability distribution vector calculation formula

The probability distribution vector refers to the English semantic dictionary, from the node corresponding to the translated English translation word to find the node corresponding to the standard English word, the node number corresponding to each word on the search path and the vector formed by the probability of the directed edge between the nodes, Its calculation formula is as follows:

Probability distribution vector _t = (1-a) × adjacency matrix × probability distribution vector _t-1 + a × probability distribution vector ₀ (6)

In the calculation formula (6), t represents the number of searches, a represents the probability of directed edges on the t-th search path, the adjacency matrix represents the word adjacency matrix in the English semantic dictionary, and the probability distribution vector ₀ represents the search start position node The probability distribution vector of , the probability distribution vector _t-1 represents the probability distribution vector of the node searched for the t-1th time, and the probability distribution vector _t represents the probability distribution vector of the node searched for the t-th time.

12. The formula for calculating the semantic similarity of the translated English translation

The semantic similarity of the translated English translation refers to the semantic similarity between the content of the translated English translation and the content of the standard English translation. Its calculation formula is as follows:

In the calculation formula (7), the one-dimensional vector of the probability distribution of the English translation means that in the English semantic dictionary, the node number corresponding to the standard English word is found from the node number corresponding to the English translation word, and the corresponding node number of each word on the search path A one-dimensional vector composed of the probability of directed edges between nodes; the one-dimensional vector of the probability distribution of the standard English translation means that in the English semantic dictionary, the node number corresponding to the translated English word is searched for the node number corresponding to the standard English word, and the search path A one-dimensional vector composed of the probability of the directed edge connected by the node at the end position; ||The one-dimensional vector of the probability distribution of the translated English translation || refers to the modulus of the one-dimensional vector of the probability distribution of the English translation; ||The one-dimensional probability distribution of the standard English translation The vector || refers to the modulus of the one-dimensional vector of the standard English translation probability distribution.

13. The formula for calculating the quality score of the translated English translation

The quality score of the translated English translation refers to the semantic similarity between the content of the translated English translation and the standard English translation. Its calculation formula is as follows:

English translation quality score = 0.5 × English translation fidelity + 0.5 × English translation semantic similarity (8)

In the calculation formula (8), the English translation fidelity score is calculated by the calculation formula (5), and the English translation semantic similarity is calculated by the calculation formula (7).

Specific steps

The processing flow charts of the English translation preprocessing module, the English translation fidelity analysis module, the English translation semantic similarity analysis module, and the English translation quality analysis result generation module of the analysis method of the present invention are as follows.

As shown in Figure 2, the processing flow of the described English translation preprocessing module is as follows:

P201 start;

P202 Read and translate English translations and standard English translations;

P203 Judging whether the number of translated English translations and standard English translations is the same, if yes, go to P204, otherwise go to P202;

P204 Segment the translated English translation and lowercase the word after the translated English translation;

P205 outputs the participle result of translation English translation and the word lowercase result of translation English translation;

P206 Statistically translates the total number of English translation words;

P207 Perform part-of-speech tagging on the word segmentation results of the translated English translation;

P208 performs phrase segmentation on the part-of-speech tagging results of the translated English translation;

P209 performs part-of-speech disambiguation on the phrase segmentation results of the translated English translation;

P210 outputs the part-of-speech tagging result of the translated English translation, the part-of-speech disambiguation result of the translated English translation, and the phrase segmentation result of the translated English translation;

P211 Segment the standard English translation and lowercase the words after the standard English translation;

P212 output the participle result of standard English translation and the word lowercase result of standard English translation;

P213 Count the total number of words in standard English batch translations;

P214 Perform part-of-speech tagging on the part-of-speech results of the standard English translation;

P215 performs phrase segmentation on the part-of-speech tagging results of the standard English translation;

P216 Perform part-of-speech disambiguation on the phrase segmentation results of the standard English translation;

P217 outputs the part-of-speech tagging result of the standard English translation, the part-of-speech disambiguation result of the standard English translation, and the phrase segmentation result of the standard English translation;

End of P218.

As shown in Figure 3, the processing flow of the English translation fidelity analysis module is as follows:

P301 start;

P302 Read in the word segmentation results of the translated English translation and the lowercase results of the translated English translation;

P303 Read the word segmentation results of the standard English translation and the word lowercase results of the standard English translation;

P304 According to the word segmentation result of the translated English translation and the word lowercase result of the translated English translation, the word segmentation result of the standard English translation and the word lowercase result of the standard English translation, generate a text pair of the translated English translation-standard English translation;

P305 reads in a set of translation English translation-standard English translation text pairs;

P306 will translate the English translation-standard English translation text pair, and each word of the Chinese translation English translation will be numbered from zero;

P307 will translate the English translation-standard English translation text pair, and each word in the standard English translation will be numbered from zero;

P308 will translate each word of the English translation in the English translation-standard English translation text pair, and perform an exact match on each word of the standard English translation in the translation English translation-standard English translation text pair;

P309 Statistically translate the English translation of the English translation-standard English translation text pair, and the English translation of the translation English translation-standard English translation text pair, the words and numbers that are accurately matched with the standard English translation;

P310 judges whether there are words that are not exactly matched in the translation English translation-standard English translation text pair, if so, turn to P308, otherwise turn to P311;

P311 will stem the English translation of the English translation-standard English translation text pair, and the English translation-standard English translation text pair that does not have an exact match in the standard English translation;

P312 Stem-matching the words in the English translation in the translated English translation-standard English translation text pair after stemming, and the word in the standard English translation in the stemmed English translation-standard English translation text pair;

P313 Statistically translate the English translation in the English translation-standard English translation text pair, and the words and numbers of the standard English translation in the translation English translation-standard English translation text pair;

P314 judges whether there are any unstemmed matching words in the text pair of translation English translation-standard English translation, if so, turn to P312, otherwise turn to P315;

P315 translates the English translation of the English translation-standard English translation text pair into the English translation, and translates the English translation-standard English translation text pair into the standard English translation for word meanings that do not have a successful match in the standard English translation;

P316 Translating words in the English translation-standard English translation text pair after paraphrasing the word meaning, and performing synonym matching on words in the standard English translation text translation in the English translation-standard English translation text pair after paraphrasing the word meaning;

P317 Statistically translate the English translation in the English translation-standard English translation text pair, and the words and numbers of the standard English translation in the English translation-standard English translation text pair that successfully match the synonyms;

P318 judges whether there are words that are not matched by synonyms in the text pair of translation English translation-standard English translation, if so, turn to P315, otherwise turn to P319;

P319 judges whether there is any uncorrected English translation-standard English translation text pair, if so, turn to P305, otherwise turn to P320;

P320 reads in the translation English translation-standard English translation text pair and the translated English translation word, and the translation English translation-standard English translation text pair and the exact matching result of the standard English translation word;

P321 reads in the translation English translation-standard English translation text pair in the translation English translation word, and the translation English translation-standard English translation text pair in the standard English translation word stem match result;

P322 read in the translation English translation-standard English translation text pair of translation English translation words, and the translation English translation-standard English translation text pair of standard English translation word synonyms matching results;

P323 marks the words that are successfully matched in the translated English translation in the text pair of translated English translation-standard English translation;

P324 The total number of words in the translated English translation in the text pair of the translated English translation-standard English translation;

P325 The total number of words in the standard English translation in the text pair of the translated English translation-standard English translation;

P326 will translate the total number of words in the translated English translation in the text pairs of the translated English translation-Standard English translation, and the total number of words in the translated English translation-Standard English translation text pairs, and the translation of the English translation-standard English translation The total number of words that are successfully matched, stem-matched, and synonymous in the translated English translation is substituted into the calculation formula (1) and calculation formula (2) of the English translation loyalty analysis module to calculate the text-to-text translation of the translated English translation-standard English translation Precision and recall of English translation word matching;

P327 uses the translated English translation set and the standard English translation set to train to obtain the values of the optimal parameters a, b, and c;

P328 Substitute the value of a, the accuracy rate of word matching between the translated English translation and the standard English translation, and the recall rate of the word matching between the translated English translation and the standard English translation into the English translation faithfulness The calculation formula (3) of the degree analysis module calculates the harmonic mean;

P329 will use the values of parameters b and c, the number of English translation phrases in the translation English translation-standard English translation text pair, and the translation English translation exact match, word stem match, and synonym match in the translation English translation-standard English translation text pair Successful total number of words, substitute into the calculation formula (4) of the English translation fidelity analysis module to calculate the penalty coefficient;

In P330, the harmonic mean and the penalty coefficient are substituted into the calculation formula (5) of the English translation fidelity analysis module to calculate the English translation fidelity;

End of P331.

As shown in Figure 4, the processing flow of the English translation semantic similarity analysis module is as follows:

P401 start;

P402 Read and translate English translation word part-of-speech disambiguation results;

P403 traversing the English translation word disambiguation results;

P404 finds and records the corresponding node number of the word in the English semantic dictionary according to the English translation word part-of-speech disambiguation result;

P405 judges whether there is any untraversed translation English translation word part-of-speech disambiguation result, if it is then turn to P403, otherwise turn to P406;

P406 Read the word part-of-speech disambiguation results of the standard English translation;

P407 Traverse standard English translation word part-of-speech disambiguation results;

P408 Find and record the corresponding node number of the word in the English semantic dictionary according to the part-of-speech disambiguation result of the standard English translation word;

P409 judges whether there are untraversed standard English translation word part-of-speech disambiguation results, if so, turn to P407, otherwise turn to P410;

P410 In the English semantic dictionary, search for the corresponding node number of the standard English word from the node number corresponding to the translated English translation word;

P411 records the node number corresponding to each word on the search path and the probability of directed edges between nodes;

P412 generates a two-dimensional vector of the probability distribution of the English translation word according to the node number corresponding to each word on the search path and the probability of the directed edge between the nodes;

P413 Calculate the two-dimensional vector of word probability distribution in the standard English translation according to the calculation formula (6) of the English translation semantic similarity analysis module;

P414 judges whether there are words not searched in the translated English translation, if so, turn to P410, otherwise turn to P415;

P415 Combine the two-dimensional vector of the probability distribution of the translated English translation words to obtain the two-dimensional vector of the probability distribution of the translated English translation;

P416 Extract the word probability value from the two-dimensional vector of the probability distribution of the translated English translation, and generate a one-dimensional vector of the probability distribution of the translated English translation;

P417 Combine the two-dimensional vector of the probability distribution of the words in the standard English translation to obtain the two-dimensional vector of the probability distribution of the standard English translation;

P418 Extract word probability values from the standard English translation probability distribution two-dimensional vector to generate a standard English translation probability distribution one-dimensional vector;

P419 calculates and outputs the semantic similarity of the English translation according to the calculation formula (7) of the translation English translation semantic similarity analysis module;

P420 ends.

As shown in Figure 5, the described translation English translation quality analysis result generation module processing flow is as follows:

P501 start;

P502 Read and translate English translation fidelity;

P503 Read and translate the semantic similarity of the English translation;

P504 calculates and outputs the English translation quality score according to the calculation formula (8) of the translation English translation quality analysis result generation module;

P505 outputs the comments of the translated English translation according to the quality score of the translated English translation;

End of P506.

Description of drawings

Fig. 1 is the overall processing flowchart of the inventive method;

Fig. 2 is the processing flowchart of the translation English translation preprocessing module of the inventive method;

Fig. 3 is the processing flowchart of the translation English translation fidelity analysis module of the inventive method;

Fig. 4 is the processing flowchart of the translation English translation semantic similarity analysis module of the inventive method;

Fig. 5 is the process flowchart of the English translation quality analysis result generating module of the method of the present invention.

detailed description

The specific implementation of the method for analyzing the translation quality of a Chinese-to-English essay of the present invention is divided into the following five steps.

Step 1: Execute the "Translation English Translation Preprocessing Module"

The English translation input in the embodiment of the present invention is drawn from the English translation questions of CET-6 for college students across the country. The Chinese translation questions, the English translation of a student’s answer, and the official standard English translation results are as follows:

Chinese translation title:

Innovation in China is booming at an unprecedented rate. In order to catch up with the world's developed countries as soon as possible in science and technology, China has greatly increased research and development funds in recent years. Chinese universities and research institutes are actively carrying out innovative research covering various high-tech fields, from big data to biochemistry, from new energy to robotics. They also collaborate with technology parks around the world to commercialize innovations. At the same time, whether in terms of products or business models, Chinese entrepreneurs are also striving to be the pioneers of innovation to adapt to the ever-changing and growing needs of domestic and foreign consumer markets.

Translate the English translation:

Innovation is progressing in an unprecedented speed in China. In order to catch up with those developed countries in the world as fast as it can in the science and technology field, China has increased funds for development research substantially in recent years. Universities and research institutions are active in China carrying out innovation researches, which cover high-technology fields such as big data, biochemistry, new energy and robots, etc. They also cooperate with science park in various regions, commercializing the research results of innovation. Meanwhile, no matter in production, business model Entrepreneurs in China are competing to be pioneers innovation to adapt to the constantly changing and increasing needs of the consumer market at home and abroad.

Standard English translation:

China's innovation is flourishing faster than ever before. In order to surpass developed countries on science and technology as soon as possible, China has sharply increased research and development fund. Chinese universities and institutes are actively doing innovative researches, covering various high field data no to biochemistry, and from newenergy to robots. They are also cooperating with science and technology parks in different places, so as to commercialize their fruits of innovation. In themeantime, to adapt to the changing foreign and domestic market, and to satisfy the growing demand, Chinese Entrepreneurs are also making pioneering efforts to innovate their products and business models.

(1) After part-of-speech tagging is performed on the translated English translation and the standard English translation, the generated part-of-speech tagging results are as follows:

Translate English translation part-of-speech tagging results:

[Innovation[innovation#NN:UN*],[#null*],is[be#VBZ*],[#null*],progressing[progress#VBG*],[#null*],in[in#IN *,in#NN*,in#RP*],[#null*],an[a#DT*],[#null*],unprecedented[unprecedented#JJ*],[#null*],speed[speed #NN:UN*,speed#VB*,speed#VBP*],[#null*],in[in#IN*,in#NN*,in#RP*],[#null*],China[China #NNP*,china#JJ*,china#NN:U*],.[.#.*]]

[In[In#NNP*,in#IN*,in#NN*,in#RP*],[#null*],order[order#NN:UN*,order#UH*,order#VB*,order #VBP*],[#null*],to[to#IN*,to#TO*],[#null*],catch[catch#NN*,catch#VB*,catch#VBP*],[# null*],up[up#IN*,up#JJ*,up#NN*,up#RP*,up#VB*,up#VBP*],[#null*],with[with#IN*, with#RP*],[#null*],those[those#DT*],[#null*],developed[develop#VBD*,develop#VBN*],[#null*],countries[country#NNS *],[#null*],in[in#IN*,in#NN*,in#RP*],[#null*],the[the#DT*],[#null*],world[world #JJ*,world#NN:UN*],[#null*],as[as#CC*,as#IN*,as#RB*],[#null*],fast[fast#JJ*,fast #NN*,fast#RP*,fast#VB*,fast#VBP*],[#null*],as[as#CC*,as#IN*,as#RB*],[#null*], it[it#PRP*],[#null*],can[can#MD*,can#NN*,can#VB*,can#VBP*],[#null*],in[in#IN*, in#NN*,in#RP*],[#null*],the[the#DT*],[#null*],science[science#NN:UN*],[#null*],and[and #CC*],[#null*],technology[technology#NN:UN*],[#null*],field[field#NN*,field#VB*,field#VBP*],,[,#, *],[#null*],China[China#NNP*,china#JJ*,china#NN:U*],[#null*],has[have#VBZ*],[#null*],increased [increase#VBD*,increase#VBN*],[#null*],funds[fund# NNS*,fund#VBZ*],[#null*],for[for#CC*,for#IN*,for#RP*],[#null*],development[development#NN:UN*],[ #null*],research[research#NN:U*,research#VB*,research#VBP*],[#null*],substantially[substantially#RB*],[#null*],in[in#IN *,in#NN*,in#RP*],[#null*],recent[recent#JJ*],[#null*],years[year#NNS*],.[.#.*]]

[Universities[university#NNS*],[#null*],and[and#CC*],[#null*],research[research#NN:U*,research#VB*,research#VBP*],[ #null*],institutions[institution#NNS*],[#null*],in[in#IN*,in#NN*,in#RP*],[#null*],China[China#NNP*, china#JJ*,china#NN:U*],[#null*],are[are#NN*,be#VBP*],[#null*],actively[actively#RB*],[#null* ],carrying[carry#VBG*],[#null*],out[out#IN*,out#NN*,out#RB*,out#RP*,out#UH*,out#VB*,out# VBP*],[#null*],innovation[innovation#NN:UN*],[#null*],researches[research#NNS*,research#VBZ*],,[,#,*],[#null *],which[which#WDT*,which#WP*],[#null*],cover[cover#NN:UN*,cover#VB*,cover#VBP*],[#null*],high- technology[high-technology#JJ*],[#null*],fields[field#NNS*,field#VBZ*],[#null*],such[such#DT*,such#PDT*],[# null*],as[as#CC*,as#IN*,as#RB*],[#null*],big[big#JJ*,big#RB*],[#null*],data[data #NN:UN*,datum#NNS*],,[,#,*],[#null*],biochemistry[biochemistry#NN:U*],,[,#,*],[#null*], new[new#JJ*],[#null*],energy[energy#NN:UN*],[#null*],and[and#CC*],[#null*],robots[robot#NNS* ],,[,#,*],[#null*],etc[etc#null*],.[. #.*],[#null*],They[they#PRP*],[#null*],also[also#RB*],[#null*],cooperate[cooperate#VB*,cooperate#VBP* ],[#null*],with[with#IN*,with#RP*],[#null*],science[science#NN:UN*],[#null*],park[park#NN*, park#VB*,park#VBP*],[#null*],in[in#IN*,in#NN*,in#RP*],[#null*],various[various#JJ*],[ #null*],regions[region#NNS*],,[,#,*],[#null*],commercializing[commercialize#VBG*],[#null*],the[the#DT*],[ #null*],research[research#NN:U*,research#VB*,research#VBP*],[#null*],results[result#NNS*,result#VBZ*],[#null*], of[of#IN*],[#null*],innovation[innovation#NN:UN*],.[.#.*]]

[Meanwhile[meanwhile#JJ*,meanwhile#NN:U*,while#RB*],,[,#,*],[#null*],no[no#DT*,no#NN*],[# null*],matter[matter#NN:UN*,matter#VB*,matter#VBP*],[#null*],in[in#IN*,in#NN*,in#RP*],[# null*],production[production#NN:UN*],[#null*],and[and#CC*],[#null*],business[business#JJ*,business#NN:UN*],[ #null*],model[model#JJ*,model#NN*,model#VB*,model#VBP*],,[,#,*],[#null*],entrepreneurs[entrepreneur#NNS*], [#null*],in[in#IN*,in#NN*,in#RP*],[#null*],China[China#NNP*,china#JJ*,china#NN:U*], [#null*],are[are#NN*,be#VBP*],[#null*],competing[compete#VBG*],[#null*],to[to#IN*,to#TO* ],[#null*],be[be#VB*],[#null*],pioneers[pioneer#NNS*,pioneer#VBZ*],[#null*],in[in#IN*,in# NN*,in#RP*],[#null*],innovation[innovation#NN:UN*],[#null*],to[to#IN*,to#TO*],[#null*], adapt[adapt#VB*,adapt#VBP*],[#null*],to[to#IN*,to#TO*],[#null*],the[the#DT*],[#null* ], constantly[constantly#RB*],[#null*],changing[change#VBG*],[#null*],and[and#CC*],[#null*],increasing[increasing#JJ* ,increasing#NN:UN*,increase#VBG*],[#null*],needs[needs#RB* ,need#NNS*,need#VBZ*],[#null*],of[of#IN*],[#null*],the[the#DT*],[#null*],consumer[consumer# NN:UN*],[#null*],market[market#NN:UN*,market#VB*,market#VBP*],[#null*],at[at#IN*,at#RP*] ,[#null*],home[home#JJ*,home#NN:UN*,home#VB*,home#VBP*],[#null*],and[and#CC*],[#null* ],abroad[abroad#JJ*,abroad#RB*],.[.#.*]]

Standard English translation part-of-speech tagging results:

[China[China#NNP*,china#JJ*,china#NN:U*],'['#null*],s[s#POS*],[#null*],innovation[innovation#NN:UN *],[#null*],is[be#VBZ*],[#null*],flourishing[flourishing#JJ*,flourish#VBG*],[#null*],faster[faster#NN*,faster #RB*,faster#RBR*,fast#JJR*],[#null*],than[than#IN*],[#null*],ever[ever#JJ*,ever#RB*,ever#RP *],[#null*],before[before#IN*,before#RP*],.[.#.*]]

[In[In#NNP*,in#IN*,in#NN*,in#RP*],[#null*],order[order#NN:UN*,order#UH*,order#VB*,order #VBP*],[#null*],to[to#IN*,to#TO*],[#null*],surpass[surpass#VB*,surpass#VBP*],[#null*],developed [develop#VBD*,develop#VBN*],[#null*],countries[country#NNS*],[#null*],on[on#IN*,on#JJ*,on#RP*], [#null*],science[science#NN:UN*],[#null*],and[and#CC*],[#null*],technology[technology#NN:UN*],[#null* ],as[as#CC*,as#IN*,as#RB*],[#null*],soon[soon#JJ*,soon#RB*],[#null*],as[as#CC *,as#IN*,as#RB*],[#null*],possible[possible#JJ*,possible#NN*],,[,#,*],[#null*],China[China# NNP*,china#JJ*,china#NN:U*],[#null*],has[have#VBZ*],[#null*],sharply[sharply#RB*],[#null*], increased[increase#VBD*,increase#VBN*],[#null*],research[research#NN:U*,research#VB*,research#VBP*],[#null*],and[and#CC *],[#null*],development[development#NN:UN*],[#null*],fund[fund#NN:UN*,fund#VB*,fund#VBP*],.[.#. *]]

[Chinese[Chinese#JJ*,Chinese#NN:U*,Chinese#NNP*],[#null*],universities[university#NNS*],[#null*],and[and#CC*],[ #null*], institutes[institute#NNS*, institute#VBZ*],[#null*],are[are#NN*,be#VBP*],[#null*],actively[actively#RB*] ,[#null*],doing[doing#NN:UN*,do#VBG*],[#null*],innovative[innovative#JJ*],[#null*],researches[research#NNS*,research #VBZ*],,[,#,*],[#null*],covering[covering#JJ*,covering#NN:UN*,cover#VBG*],[#null*],various[various#JJ *],[#null*],fields[field#NNS*,field#VBZ*],[#null*],of[of#IN*],[#null*],high[high#JJ*,high #NN*,high#RP*],[#null*],technology[technology#NN:UN*],,[,#,*],[#null*],from[from#IN*,from#RP *],[#null*],big[big#JJ*,big#RB*],[#null*],data[data#NN:UN*,datum#NNS*],[#null*],to [to#IN*,to#TO*],[#null*],biochemistry[biochemistry#NN:U*],[,#,*],[#null*],and[and#CC*], [#null*],from[from#IN*,from#RP*],[#null*],new[new#JJ*],[#null*],energy[energy#NN:UN*],[ #null*],to[to#IN*,to#TO*],[#null*],robots[robot#NNS*],.[.#.*]]

[They[they#PRP*],[#null*],are[are#NN*,be#VBP*],[#null*],also[also#RB*],[#null*],cooperating[ cooperate#VBG*],[#null*],with[with#IN*,with#RP*],[#null*],science[science#NN:UN*],[#null*],and[and #CC*],[#null*],technology[technology#NN:UN*],[#null*],parks[park#NNS*,park#VBZ*],[#null*],in[in# IN*,in#NN*,in#RP*],[#null*],different[different#JJ*],[#null*],places[place#NNS*,place#VBZ*],,[, #,*],[#null*],so[so#CC*,so#JJ*,so#NN:U*],[#null*],as[as#CC*,as#IN*,as #RB*],[#null*],to[to#IN*,to#TO*],[#null*],commercialize[commercialize#VB*,commercialize#VBP*],[#null*],their [their#PRP$*],[#null*],fruits[fruit#NNS*,fruit#VBZ*],[#null*],of[of#IN*],[#null*],innovation[innovation #NN:UN*],.[.#.*]]

[In[In#NNP*,in#IN*,in#NN*,in#RP*],[#null*],the[the#DT*],[#null*],meantime[meantime#JJ* ,meantime#NN:U*],,[,#,*],[#null*],to[to#IN*,to#TO*],[#null*],adapt[adapt#VB*,adapt #VBP*],[#null*],to[to#IN*,to#TO*],[#null*],the[the#DT*],[#null*],changing[change#VBG* ],[#null*],foreign[foreign#JJ*],[#null*],and[and#CC*],[#null*],domestic[domestic#JJ*,domestic#NN*],[ #null*],market[market#NN:UN*,market#VB*,market#VBP*],,[,#,*],[#null*],and[and#CC*],[#null *],to[to#IN*,to#TO*],[#null*],satisfy[satisfy#VB*,satisfy#VBP*],[#null*],the[the#DT*],[ #null*],growing[growing#NN:UN*,grow#VBG*],[#null*],demand[demand#NN:UN*,demand#VB*,demand#VBP*],,[,# ,*],[#null*],Chinese[Chinese#JJ*,Chinese#NN:U*,Chinese#NNP*],[#null*],entrepreneurs[entrepreneur#NNS*],[#null*], are[are#NN*,be#VBP*],[#null*],also[also#RB*],[#null*],making[making#NN:UN*,make#VBG*],[# null*],pioneering[pioneer#VBG*],[#null*],efforts[effort#NNS*],[#null*],to[to#IN*,to#TO*],[#null*] ,innovate[innovate#VB*,innovate#VBP*],[#null*],their[their#PRP$*],[#null*] ,products[product#NNS*],[#null*],and[and#CC*],[#null*],business[business#JJ*,business#NN:UN*],[#null*], models[model#NNS*,model#VBZ*],.[.#.*]]

(2) After performing phrase segmentation on the translated English translation and the standard English translation, the generated phrase segmentation results are as follows:

Translating the English translation phrase chopping results:

/Innovation/innovation#NN:UN*,B-NP-singular|E-NP-singular/,/#null*/,is/be#VBZ*,B-VP/,/#null*/,progressing/progress #VBG*,I-VP/,/#null*/,in/in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,an/a#DT*, B-NP-singular/,/#null*/,unprecedented/unprecedented#JJ*,I-NP-singular/,/#null*/,speed/speed#NN:UN*,speed#VB*,speed#VBP *,E-NP-singular/,/#null*/,in/in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,China/China#NNP*, china#JJ*,china#NN:U*,B-NP-singular|E-NP-singular/,./.#.*,O//

/In/In#NNP*,in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,order/order#NN:UN*,order#UH*,order# VB*,order#VBP*,B-NP-singular|E-NP-singular/,/#null*/,to/to#IN*,to#TO*,B-VP/,/#null*/, catch/catch#NN*,catch#VB*,catch#VBP*,I-VP/,/#null*/,up/up#IN*,up#JJ*,up#NN*,up#RP*, up#VB*,up#VBP*,B-PRT/,/#null*/,with/with#IN*,with#RP*,B-PP/,/#null*/,those/those#DT* ,B-NP-plural/,/#null*/,developed/develop#VBD*,develop#VBN*,I-NP-plural/,/#null*/,countries/country#NNS*,E-NP- plural/,/#null*/,in/in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,the/the#DT*,B-NP-singular/ ,/#null*/,world/world#JJ*,world#NN:UN*,E-NP-singular/,/#null*/,as/as#CC*,as#IN*,as#RB* ,B-ADVP/,/#null*/,fast/fast#JJ*,fast#NN*,fast#RP*,fast#VB*,fast#VBP*,I-ADVP/,/#null*/, as/as#CC*,as#IN*,as#RB*,B-SBAR/,/#null*/,it/it#PRP*,B-NP-singular|E-NP-singular/,/# null*/,can/can#MD*,can#NN*,can#VB*,can#VBP*,B-VP/,/#null*/,in/in#IN*,in#NN*,in #RP*,B-PP/,/#null*/,the/the#DT*,B-NP-singular/,/#null*/,science/science#NN:UN*,I-NP-singular/ ,/#null*/,and/and#CC*,I-NP-singular/,/#null*/,te chnology/technology#NN:UN*,I-NP-singular/,/#null*/,field/field#NN*,field#VB*,field#VBP*,E-NP-singular/,,/,# ,*,O/,/#null*/,China/China#NNP*,china#JJ*,china#NN:U*,B-NP-singular|E-NP-singular/,/#null*/, has/have#VBZ*,B-VP/,/#null*/,increased/increase#VBD*,increase#VBN*,I-VP/,/#null*/,funds/fund#NNS*,fund# VBZ*,B-NP-plural|E-NP-plural/,/#null*/,for/for#CC*,for#IN*,for#RP*,B-PP/,/#null*/, development/development#NN:UN*,B-NP-singular/,/#null*/,research/research#NN:U*,research#VB*,research#VBP*,E-NP-singular/,/# null*/,substantially/substantially#RB*,B-ADVP/,/#null*/,in/in#IN*,in#NN*,in#RP*,B-PP/,/#null*/, recent/recent#JJ*,B-NP-plural/,/#null*/,years/year#NNS*,E-NP-plural/,./.#.*,O//

/Universities/university#NNS*,B-NP-plural/,/#null*/,and/and#CC*,I-NP-plural/,/#null*/,research/research#NN:U*, research#VB*,research#VBP*,I-NP-plural/,/#null*/,institutions/institution#NNS*,E-NP-plural/,/#null*/,in/in#IN*, in#NN*,in#RP*,B-PP/,/#null*/,China/China#NNP*,china#JJ*,china#NN:U*,B-NP-singular|E-NP- singular/,/#null*/,are/are#NN*,be#VBP*,B-VP/,/#null*/,actively/actively#RB*,I-VP/,/#null*/, carrying/carry#VBG*,I-VP/,/#null*/,out/out#IN*,out#NN*,out#RB*,out#RP*,out#UH*,out#VB*, out#VBP*,B-PRT/,/#null*/,innovation/innovation#NN:UN*,B-NP-plural/,/#null*/,researches/research#NNS*,research#VBZ*, E-NP-plural/,,/,#,*,O/,/#null*/,which/which#WDT*,which#WP*,B-NP-singular|E-NP-singular/,/# null*/,cover/cover#NN:UN*,cover#VB*,cover#VBP*,B-VP/,/#null*/,high-technology/high-technology#JJ*,B-NP-plural /,/#null*/,fields/field#NNS*,field#VBZ*,E-NP-plural/,/#null*/,such/such#DT*,such#PDT*,B-PP/, /#null*/,as/as#CC*,as#IN*,as#RB*,I-PP/,/#null*/,big/big#JJ*,big#RB*,B-NP- plural/,/#null*/,data/data#NN:UN*,d atum#NNS*,E-NP-plural/,,/,#,*,O/,/#null*/,biochemistry/biochemistry#NN:U*,B-NP-singular|E-NP-singular/, ,/,#,*,O/,/#null*/,new/new#JJ*,B-NP-plural/,/#null*/,energy/energy#NN:UN*,I-NP-plural /,/#null*/,and/and#CC*,I-NP-plural/,/#null*/,robots/robot#NNS*,E-NP-plural/,,/,#,*,O /,/#null*/,etc/etc#null*,B-NP-singular|E-NP-singular/,./.#.*,O/,/#null*/,They/they#PRP* ,B-NP-singular|E-NP-singular/,/#null*/,also/also#RB*,B-ADVP/,/#null*/,cooperate/cooperate#VB*,cooperate#VBP*, B-VP/,/#null*/,with/with#IN*,with#RP*,B-PP/,/#null*/,science/science#NN:UN*,B-NP-singular/, /#null*/,park/park#NN*,park#VB*,park#VBP*,E-NP-singular/,/#null*/,in/in#IN*,in#NN*,in# RP*,B-PP/,/#null*/,various/various#JJ*,B-NP-plural/,/#null*/,regions/region#NNS*,E-NP-plural/,,/ ,#,*,O/,/#null*/,commercializing/commercialize#VBG*,B-VP/,/#null*/,the/the#DT*,B-NP-plural/,/#null* /,research/research#NN:U*,research#VB*,research#VBP*,I-NP-plural/,/#null*/,results/result#NNS*,result#VBZ*,E-NP- plural/,/#null*/,of/of#IN*,B-PP/,/#null*/,innova tion/innovation#NN:UN*,B-NP-singular|E-NP-singular/,./.#.*,O//

/Meanwhile/meanwhile#JJ*,meanwhile#NN:U*,while#RB*,B-ADVP/,,/,#,*,O/,/#null*/,no/no#DT*,no# NN*,B-NP-singular/,/#null*/,matter/matter#NN:UN*,matter#VB*,matter#VBP*,E-NP-singular/,/#null*/,in/ in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,production/production#NN:UN*,B-NP-singular/,/#null*/,and/ and#CC*,I-NP-singular/,/#null*/,business/business#JJ*,business#NN:UN*,I-NP-singular/,/#null*/,model/model#JJ *,model#NN*,model#VB*,model#VBP*,E-NP-singular/,,/,#,*,O/,/#null*/,entrepreneurs/entrepreneur#NNS*,B-NP -plural|E-NP-plural/,/#null*/,in/in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,China/China#NNP* ,china#JJ*,china#NN:U*,B-NP-singular|E-NP-singular/,/#null*/,are/are#NN*,be#VBP*,B-VP/,/ #null*/,competing/compete#VBG*,I-VP/,/#null*/,to/to#IN*,to#TO*,I-VP/,/#null*/,be/be# VB*,I-VP/,/#null*/,pioneers/pioneer#NNS*,pioneer#VBZ*,B-NP-plural|E-NP-plural/,/#null*/,in/in#IN *,in#NN*,in#RP*,B-PP/,/#null*/,innovation/innovation#NN:UN*,B-NP-singular|E-NP-singular/,/#null*/ ,to/to#IN*,to#TO*,B-VP/,/#null*/,adapt/adapt#VB*,adapt pt#VBP*,I-VP/,/#null*/,to/to#IN*,to#TO*,B-PP/,/#null*/,the/the#DT*,B-NP- singular/,/#null*/,constantly/constantly#RB*,I-NP-singular/,/#null*/,changing/change#VBG*,E-NP-singular/,/#null*/,and /and#CC*,O/,/#null*/,increasing/increasing#JJ*,increasing#NN:UN*,increase#VBG*,B-VP/,/#null*/,needs/needs#RB *,need#NNS*,need#VBZ*,B-NP-plural|E-NP-plural/,/#null*/,of/of#IN*,B-PP/,/#null*/,the /the#DT*,B-NP-singular/,/#null*/,consumer/consumer#NN:UN*,I-NP-singular/,/#null*/,market/market#NN:UN*, market#VB*,market#VBP*,E-NP-singular/,/#null*/,at/at#IN*,at#RP*,B-PP/,/#null*/,home/home# JJ*,home#NN:UN*,home#VB*,home#VBP*,B-NP-singular|E-NP-singular/,/#null*/,and/and#CC*,O/,/ #null*/,abroad/abroad#JJ*,abroad#RB*,B-ADVP/,./.#.*,O//

Standard English translation phrase segmentation results:

/China/China#NNP*,china#JJ*,china#NN:U*,B-NP-singular|E-NP-singular/,'/'#null*/,s/s#POS*/,/ #null*/,innovation/innovation#NN:UN*,E-NP-singular/,/#null*/,is/be#VBZ*,B-VP/,/#null*/,flourishing/flourishing#JJ *,flourish#VBG*,I-VP/,/#null*/,faster/faster#NN*,faster#RB*,faster#RBR*,fast#JJR*,B-ADVP/,/#null*/ ,than/than#IN*,B-PP/,/#null*/,ever/ever#JJ*,ever#RB*,ever#RP*,B-ADVP/,/#null*/,before/before #IN*,before#RP*,I-ADVP/,./.#.*,O//

/In/In#NNP*,in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,order/order#NN:UN*,order#UH*,order# VB*,order#VBP*,B-NP-singular|E-NP-singular/,/#null*/,to/to#IN*,to#TO*,B-VP/,/#null*/, surpass/surpass#VB*,surpass#VBP*,I-VP/,/#null*/,developed/develop#VBD*,develop#VBN*,B-NP-plural/,/#null*/,countries/ country#NNS*,E-NP-plural/,/#null*/,on/on#IN*,on#JJ*,on#RP*,B-PP/,/#null*/,science/science# NN:UN*,B-NP-singular/,/#null*/,and/and#CC*,I-NP-singular/,/#null*/,technology/technology#NN:UN*,E-NP -singular/,/#null*/,as/as#CC*,as#IN*,as#RB*,B-ADVP/,/#null*/,soon/soon#JJ*,soon#RB*, I-ADVP/,/#null*/,as/as#CC*,as#IN*,as#RB*,B-PP/,/#null*/,possible/possible#JJ*,possible#NN* ,B-ADJP/,,/,#,*,O/,/#null*/,China/China#NNP*,china#JJ*,china#NN:U*,B-NP-singular|E-NP -singular/,/#null*/,has/have#VBZ*,B-VP/,/#null*/,sharply/sharply#RB*,I-VP/,/#null*/,increased/increase# VBD*,increase#VBN*,I-VP/,/#null*/,research/research#NN:U*,research#VB*,research#VBP*,B-NP-singular/,/#null*/ ,and/and#CC*,I-NP-singular/,/#null*/,development/development ment#NN:UN*,I-NP-singular/,/#null*/,fund/fund#NN:UN*,fund#VB*,fund#VBP*,E-NP-singular/,./.# .*,O//

/Chinese/Chinese#JJ*,Chinese#NN:U*,Chinese#NNP*,B-NP-plural/,/#null*/,universities/university#NNS*,I-NP-plural/,/#null */,and/and#CC*,I-NP-plural/,/#null*/,institutes/institute#NNS*,institute#VBZ*,E-NP-plural/,/#null*/,are/ are#NN*,be#VBP*,B-VP/,/#null*/,actively/actively#RB*,I-VP/,/#null*/,doing/doing#NN:UN*,do# VBG*,I-VP/,/#null*/,innovative/innovative#JJ*,B-NP-plural/,/#null*/,researches/research#NNS*,research#VBZ*,E-NP- plural/,,/,#,*,O/,/#null*/,covering/covering#JJ*,covering#NN:UN*,cover#VBG*,B-VP/,/#null*/,various /various#JJ*,B-NP-plural/,/#null*/,fields/field#NNS*,field#VBZ*,E-NP-plural/,/#null*/,of/of#IN* ,B-PP/,/#null*/,high/high#JJ*,high#NN*,high#RP*,B-NP-singular/,/#null*/,technology/technology#NN:UN* ,E-NP-singular/,,/,#,*,O/,/#null*/,from/from#IN*,from#RP*,B-PP/,/#null*/,big/big #JJ*,big#RB*,B-NP-plural/,/#null*/,data/data#NN:UN*,datum#NNS*,E-NP-plural/,/#null*/,to /to#IN*,to#TO*,B-VP/,/#null*/,biochemistry/biochemistry#NN:U*,I-VP/,,/,#,*,O/,/#null* /,and/and#CC*,O/,/#null*/,from/ from#IN*,from#RP*,B-PP/,/#null*/,new/new#JJ*,B-NP-singular/,/#null*/,energy/energy#NN:UN*, E-NP-singular/,/#null*/,to/to#IN*,to#TO*,B-PP/,/#null*/,robots/robot#NNS*,B-NP-plural|E -NP-plural/,./.#.*,O//

/They/they#PRP*,B-NP-singular|E-NP-singular/,/#null*/,are/are#NN*,be#VBP*,B-VP/,/#null*/, also/also#RB*,I-VP/,/#null*/,cooperating/cooperate#VBG*,I-VP/,/#null*/,with/with#IN*,with#RP*,B- PP/,/#null*/,science/science#NN:UN*,B-NP-plural/,/#null*/,and/and#CC*,I-NP-plural/,/#null*/ ,technology/technology#NN:UN*,I-NP-plural/,/#null*/,parks/park#NNS*,park#VBZ*,E-NP-plural/,/#null*/,in/ in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,different/different#JJ*,B-NP-plural/,/#null*/,places/place# NNS*,place#VBZ*,E-NP-plural/,,/,#,*,O/,/#null*/,so/so#CC*,so#JJ*,so#NN:U*, O/,/#null*/,as/as#CC*,as#IN*,as#RB*,O/,/#null*/,to/to#IN*,to#TO*,B-VP /,/#null*/,commercialize/commercialize#VB*,commercialize#VBP*,I-VP/,/#null*/,their/their#PRP$*,B-NP-plural/,/#null* /,fruits/fruit#NNS*,fruit#VBZ*,E-NP-plural/,/#null*/,of/of#IN*,B-PP/,/#null*/,innovation/innovation#NN :UN*,B-NP-singular|E-NP-singular/,./.#.*,O//

/In/In#NNP*,in#IN*,in#NN*,in#RP*,B-PP/,/#null*/,the/the#DT*,B-NP-singular/,/# null*/,meantime/meantime#JJ*,meantime#NN:U*,E-NP-singular/,,/,#,*,O/,/#null*/,to/to#IN*,to# TO*,B-VP/,/#null*/,adapt/adapt#VB*,adapt#VBP*,I-VP/,/#null*/,to/to#IN*,to#TO*,B -PP/,/#null*/,the/the#DT*,B-NP-singular/,/#null*/,changing/change#VBG*,I-NP-singular/,/#null*/, foreign/foreign#JJ*,I-NP-singular/,/#null*/,and/and#CC*,I-NP-singular/,/#null*/,domestic/domestic#JJ*,domestic#NN *,I-NP-singular/,/#null*/,market/market#NN:UN*,market#VB*,market#VBP*,E-NP-singular/,,/,#,*,O/ ,/#null*/,and/and#CC*,O/,/#null*/,to/to#IN*,to#TO*,B-VP/,/#null*/,satisfy/satisfy# VB*,satisfy#VBP*,I-VP/,/#null*/,the/the#DT*,B-NP-singular/,/#null*/,growing/growing#NN:UN*,grow# VBG*,I-NP-singular/,/#null*/,demand/demand#NN:UN*,demand#VB*,demand#VBP*,E-NP-singular/,,/,#,*,O /,/#null*/,Chinese/Chinese#JJ*,Chinese#NN:U*,Chinese#NNP*,B-NP-plural/,/#null*/,entrepreneurs/entrepreneur#NNS*,E-NP -plural/,/#null*/,are/are#NN*,be#VBP*,B-VP/,/#null*/,also/als o#RB*,I-VP/,/#null*/,making/making#NN:UN*,make#VBG*,I-VP/,/#null*/,pioneering/pioneer#VBG*,B- NP-plural/,/#null*/,efforts/effort#NNS*,E-NP-plural/,/#null*/,to/to#IN*,to#TO*,B-VP/,/# null*/,innovate/innovate#VB*,innovate#VBP*,I-VP/,/#null*/,their/their#PRP$*,B-NP-plural/,/#null*/,products/ product#NNS*,I-NP-plural/,/#null*/,and/and#CC*,I-NP-plural/,/#null*/,business/business#JJ*,business#NN:UN *,I-NP-plural/,/#null*/,models/model#NNS*,model#VBZ*,E-NP-plural/,./.#.*,O//

(3) After performing part-of-speech disambiguation on the translated English translation and the standard English translation, the generated part-of-speech disambiguation results are as follows:

Translate English translation part-of-speech disambiguation results:

<S>Innovation[innovation#NN:UN,B-NP-singular|E-NP-singular]is[be#VBZ,B-VP]progressing[progress#VBG,I-VP]in[in#IN,B -PP]an[a#DT,B-NP-singular]unprecedented[unprecedented#JJ,I-NP-singular]speed[speed#NN:UN,E-NP-singular]in[in#IN,B-PP ]China[China#NNP,B-NP-singular|E-NP-singular].[.#.,<#S>,O]

<S>In[in#IN,B-PP]order[order#NN:UN,B-NP-singular|E-NP-singular]to[to#IN,to#TO,B-VP]catch[catch #NN,catch#VB,catch#VBP,I-VP]up[up#IN,up#JJ,up#NN,up#RP,up#VB,up#VBP,B-PRT]with[with#IN ,with#RP,B-PP]those[those#DT,B-NP-plural]developed[develop#VBD,develop#VBN,I-NP-plural]countries[country#NNS,E-NP-plural]in [in#IN,B-PP]the[the#DT,B-NP-singular]world[world#NN:UN,E-NP-singular]as[as#RB,B-ADVP]fast[fast#JJ ,fast#NN,fast#RP,fast#VB,fast#VBP,I-ADVP]as[as#RB,B-SBAR]it[it#PRP,B-NP-singular|E-NP-singular]can [can#MD,B-VP]in[in#IN,B-PP]the[the#DT,B-NP-singular]science[science#NN:UN,I-NP-singular]and[and#CC ,I-NP-singular]technology[technology#NN:UN,I-NP-singular]field[field#NN,field#VB,field#VBP,E-NP-singular],[,#,,O]China [China#NNP,B-NP-singular|E-NP-singular]has[have#VBZ,B-VP]increased[increase#VBN,I-VP]funds[fund#NNS,fund#VBZ,B-NP -plural|E-NP-plural]for[for#CC,for#IN,for#RP,B-PP]development[development#NN:UN,B-NP-singular]research[research#NN:U,research #VB,research#VBP,E-NP-singular]substantially[sub constantly#RB,B-ADVP]in[in#IN,in#NN,in#RP,B-PP]recent[recent#JJ,B-NP-plural]years[year#NNS,E-NP-plural] .[.#.,<#S>,O]

<S>Universities[university#NNS,B-NP-plural]and[and#CC,I-NP-plural]research[research#NN:U,research#VB,research#VBP,I-NP-plural]institutions [institution#NNS,E-NP-plural]in[in#IN,B-PP]China[China#NNP,B-NP-singular|E-NP-singular]are[be#VBP,B-VP]actively [actively#RB,I-VP]carrying[carry#VBG,I-VP]out[out#IN,out#NN,out#RB,out#RP,out#UH,B-PRT]innovation[innovation#NN :UN,B-NP-plural]researches[research#NNS,research#VBZ,E-NP-plural],[,#,,O]which[which#WDT,B-NP-singular|E-NP-singular ]cover[cover#NN:UN,cover#VB,cover#VBP,B-VP]high-technology[high-technology#JJ,B-NP-plural]fields[field#NNS,E-NP-plural]such [such#DT,B-PP]as[as#CC,as#IN,as#RB,I-PP]big[big#JJ,big#RB,B-NP-plural]data[data#NN:UN ,datum#NNS,E-NP-plural],[,#,,O]biochemistry[biochemistry#NN:U,B-NP-singular|E-NP-singular],[,#,,O]new[new #JJ,B-NP-plural]energy[energy#NN:UN,I-NP-plural]and[and#CC,I-NP-plural]robots[robot#NNS,E-NP-plural],[, #,,O]etc[etc#null,B-NP-singular|E-NP-singular].[.#.,O]They[they#PRP,B-NP-singular|E-NP-singular]also [also #RB ,B-ADVP]cooperate[cooperate#VB,cooperate#VBP,B-VP]with[with#IN,with#RP,B-PP]science[science#NN:UN,B-NP-singular]park[park #NN,park#VB,park#VBP,E-NP-singular]in[in#IN,in#NN,in#RP,B-PP]various[various#JJ,B-NP-plural]regions[region #NNS,E-NP-plural],[,#,,O]commercializing[commercialize#VBG,B-VP]the[the#DT,B-NP-plural]research[research#NN:U,I-NP -plural]results[result#NNS,result#VBZ,E-NP-plural]of[of#IN,B-PP]innovation[innovation#NN:UN,B-NP-singular|E-NP-singular]. [.#.,<#S>,O]

<S>Meanwhile[meanwhile#JJ,meanwhile#NN:U,while#RB,B-ADVP],[,#,,O]no[no#DT,B-NP-singular]matter[matter#NN:UN ,E-NP-singular]in[in#IN,B-PP]production[production#NN:UN,B-NP-singular]and[and#CC,I-NP-singular]business[business#JJ,I -NP-singular]model[model#JJ,model#NN,model#VB,model#VBP,E-NP-singular],[,#,,O]entrepreneurs[entrepreneur#NNS,B-NP-plural|E -NP-plural]in[in#IN,B-PP]China[China#NNP,B-NP-singular|E-NP-singular]are[be#VBP,B-VP]competing[compete#VBG,I -VP]to[to#IN,to#TO,I-VP]be[be#VB,I-VP]pioneers[pioneer#NNS,B-NP-plural|E-NP-plural]in[in#IN ,B-PP] innovation[innovation#NN:UN,B-NP-singular|E-NP-singular]to[to#IN,to#TO,B-VP]adapt[adapt#VB,adapt#VBP,I -VP]to[to#IN,to#TO,B-PP]the[the#DT,B-NP-singular]constantly[constantly#RB,I-NP-singular]changing[change#VBG,E-NP -singular]and[and#CC,O]increasing[increasing#JJ,increasing#NN:UN,increase#VBG,B-VP]needs[need#NNS,B-NP-plural|E-NP-plural]of [of#IN,B-PP]the[the#DT,B-NP-singular]consumer[consumer#NN:UN,I-NP-singular]market[market#NN:UN,E-NP-singular] ar]at[at#IN,at#RP,B-PP]home[home#NN:UN,B-NP-singular|E-NP-singular]and[and#CC,O]abroad[abroad#JJ, abroad#RB,B-ADVP].[.#.,<#S>,O]

Standard English translation part-of-speech disambiguation results:

<S>

China[China#NNP,china#JJ,china#NN:U,B-NP-singular|E-NP-singular]'['#POS]s[s#POS]innovation[innovation#NN:UN,E- NP-singular]is[be#VBZ,B-VP]flourishing[flourish#VBG,I-VP]faster[faster#RBR,fast#JJR,B-ADVP]than[than#IN,B-PP]ever[ ever#JJ,ever#RB,ever#RP,B-ADVP]before[before#IN,before#RP,I-ADVP].[.#.,<#S>,O]

<S>In[in#IN,B-PP]order[order#NN:UN,B-NP-singular|E-NP-singular]to[to#IN,to#TO,B-VP]surpass[surpass #VB,surpass#VBP,I-VP]developed[develop#VBD,develop#VBN,B-NP-plural]countries[country#NNS,E-NP-plural]on[on#IN,on#JJ,on #RP,B-PP]science[science#NN:UN,B-NP-singular]and[and#CC,I-NP-singular]technology[technology#NN:UN,E-NP-singular]as[as #RB,B-ADVP]soon[soon#JJ,soon#RB,I-ADVP]as[as#RB,B-PP]possible[possible#JJ,possible#NN,B-ADJP],[,#, ,O]China[China#NNP,B-NP-singular|E-NP-singular]has[have#VBZ,B-VP]sharply[sharply#RB,I-VP]increased[increase#VBN,I-VP ]research[research#NN:U,B-NP-singular]and[and#CC,I-NP-singular]development[development#NN:UN,I-NP-singular]fund[fund#NN:UN,fund #VB,fund#VBP,E-NP-singular].[.#.,<#S>,O]

<S>Chinese[Chinese#JJ,B-NP-plural]universities[university#NNS,I-NP-plural]and[and#CC,I-NP-plural]institutes[institute#NNS,E-NP-plural] ]are[be#VBP,B-VP]actively[actively#RB,I-VP]doing[doing#NN:UN,do#VBG,I-VP]innovative[innovative#JJ,B-NP-plural]researches [research#NNS,E-NP-plural],[,#,,O]covering[covering#JJ,covering#NN:UN,cover#VBG,B-VP]various[various#JJ,B-NP-plural ]fields[field#NNS,E-NP-plural]of[of#IN,B-PP]high[high#JJ,high#NN,high#RP,B-NP-singular]technology[technology#NN:UN ,E-NP-singular],[,#,,O]from[from#IN,from#RP,B-PP]big[big#JJ,big#RB,B-NP-plural]data[data#NN :UN,datum#NNS,E-NP-plural]to[to#IN,to#TO,B-VP]biochemistry[biochemistry#NN:U,I-VP],[,#,,O]and[and #CC,O]from[from#IN,from#RP,B-PP]new[new#JJ,B-NP-singular]energy[energy#NN:UN,E-NP-singular]to[to#IN ,to#TO,B-PP]robots[robot#NNS,B-NP-plural|E-NP-plural].[.#.,<#S>,O]

<S>They[they#PRP,B-NP-singular|E-NP-singular]are[be#VBP,B-VP]also[also#RB,I-VP]cooperating[cooperate#VBG,I-VP ]with[with#IN,with#RP,B-PP]science[science#NN:UN,B-NP-plural]and[and#CC,I-NP-plural]technology[technology#NN:UN,I -NP-plural]parks[park#NNS,park#VBZ,E-NP-plural]in[in#IN,in#NN,in#RP,B-PP]different[different#JJ,B-NP-plural] ]places[place#NNS,E-NP-plural],[,#,,O]so[so#RB,O]as[as#CC,as#IN,O]to[to#TO,B-VP ]commercialize[commercialize#VB,I-VP]their[their#PRP$,B-NP-plural]fruits[fruit#NNS,E-NP-plural]of[of#IN,B-PP]innovation[innovation# NN:UN,B-NP-singular|E-NP-singular].[.#.,<#S>,O]

<S>In[in#IN,B-PP]the[the#DT,B-NP-singular]meantime[meantime#JJ,meantime#NN:U,E-NP-singular],[,#,,O ]to[to#IN,to#TO,B-VP]adapt[the#VB,adapt#VBP,I-VP]to[to#IN,to#TO,B-PP]the[the#DT,B -NP-singular]changing[change#VBG,I-NP-singular]foreign[foreign#JJ,I-NP-singular]and[and#CC,I-NP-singular]domestic[domestic#JJ,I-NP -singular]market[market#NN:UN,E-NP-singular],[,#,,O]and[and#CC,O]to[to#TO,B-VP]satisfy[satisfy#VB,I -VP]the[the#DT,B-NP-singular]growing[growing#NN:UN,grow#VBG,I-NP-singular]demand[demand#NN:UN,E-NP-singular],[, #,,O]Chinese[Chinese#JJ,B-NP-plural]entrepreneurs[entrepreneur#NNS,E-NP-plural]are[be#VBP,B-VP]also[also#RB,I-VP]making [making#NN:UN,make#VBG,I-VP]pioneering[pioneer#VBG,B-NP-plural]efforts[effort#NNS,E-NP-plural]to[to#TO,B-VP]innovate [innovate#VB,I-VP]their[their#PRP$,B-NP-plural]products[product#NNS,I-NP-plural]and[and#CC,I-NP-plural]business[business# JJ,I-NP-plural]models[model#NNS,E-NP-plural].[.#.,<#S>,O]

Step 2: Execute the "Translation English Translation Loyalty Analysis Module"

The English translation fidelity analysis module uses the English translation generated in the first step and the word segmentation results of the standard English translation preprocessing module, and then performs exact matching, word stem matching, and synonym matching on the words in the translated English translation, and Record the successfully matched words and their numbers, and finally use the exact match weight, stem match weight, synonym match weight generated by the training English text set and the total number of successfully matched words in the translated English translation to calculate the fidelity of the translated English translation. The English translation faithfulness analysis result of this embodiment is as follows:

Mark the correct English word in the translated English translation:

Innovation is progressing in an unprecedented speed in China . In order to catch up with those developed countries in the world as fast as it can in the science and technology field, China has increased funds for development research substantially in recent years. Universities and research institutions in China are actively carrying out innovation researches , which cover high - technology fields such as big data , biochemistry , new energy and robots ,etc. They also cooperate with science park in various regions, commercializing the research results of innovation , Meanwhile production and business model , entrepreneurs in China are competing to be pioneers in innovation to adapt to the constantly changing and increasing needs of the consumer market at home and abroad.

English translation fidelity: 63 points.

Step 3: Execute the "Semantic Similarity Analysis Module of Translated English Translation"

The English translation semantic similarity analysis module is to use the translation English translation generated in the first step and the standard English translation part-of-speech disambiguation results to find and record the node numbers corresponding to the words in the English semantic dictionary according to the part-of-speech disambiguation results. , find the node number corresponding to the standard English word from the node number corresponding to the translated English word, record the node number corresponding to each word on the search path and the probability of the directed edge between nodes, according to each word on the search path The corresponding node number and the probability of the directed edge between the nodes generate a one-dimensional vector of the probability distribution of the translated English translation and a one-dimensional vector of the probability distribution of the standard English translation. According to the one-dimensional vector of the probability distribution of the translated English translation and the one-dimensional probability distribution of the standard English translation Vector computes and outputs the semantic similarity of the translated English text. The semantic similarity analysis results of the translated English translation of the present embodiment are as follows:

Translate the English translation probability distribution two-dimensional vector:

Node number = 2156 Probability value = 3.0124842E-4

Node number = 4370 Probability value = 5.800353E-4

node number = 6845 probability value = 6.6988135E-4

Node number = 100600 Probability value = 2.8010816E-4

Node number = 10177 Probability value = 2.602892E-4

node number = 46299 probability value = 0.021831261

Node number = 15673 Probability value = 2.3386389E-4

node number = 105382 probability value = 0.0012974822

Node number = 54185 Probability value = 2.3650641E-4

Node number = 17634 Probability value = 5.919267E-4

node number = 20446 probability value = 0.00863843

node number = 14347 probability value = 4.175197E-4

Node number = 20828 Probability value = 6.5931125E-4

node number = 19141 probability value = 2.867145E-4

node number = 26691 probability value = 9.737723e-4

node number = 102174 probability value = 0.0026993442

Node number = 28336 Probability value = 2.3122136E-4

node number = 29957 probability value = 7.412297e-4

node number = 22319 probability value = 2.4179148E-4

node number = 43079 probability value = 0.0014851018

node number = 32419 probability value = 0.00199511

node number = 35642 probability value = 2.8142944E-4

node number = 4743 probability value = 3.0917602E-4

node number = 40963 probability value = 6.0513936E-4

node number = 89379 probability value = 2.4311275E-4

Node number = 44605 Probability value = 2.668955E-4

node number = 55839 probability value = 3.0124842E-4

node number = 46581 probability value = 4.294111E-4

node number = 48312 probability value = 2.906783e-4

node number = 52569 probability value = 0.018896732

Node number = 53413 Probability value = 4.294111E-4

Node number = 97348 Prob value = 4.2544733E-4

node number = 41450 probability value = 0.002328069

node number = 58665 probability value = 2.8142944E-4

Node number = 30269 Probability value = 3.052122E-4

Node number = 59301 Probability value = 5.1132956E-4

Node number = 62168 Probability value = 4.2676856E-4

Node number = 62590 Probability value = 6.276009E-4

Node number = 65360 Probability value = 2.4707653E-4

Node number = 6306 Probability value = 0.002806367

node number = 67688 probability value = 3.4881395E-4

node number = 19078 probability value = 4.6640655E-4

node number = 71778 probability value = 3.5013523e-4

node number = 72009 probability value = 4.2676856E-4

Node number = 6295 Probability value = 3.3428005E-4

Node number = 19068 Probability value = 2.3386389E-4

node number = 113288 probability value = 4.928318E-4

node number = 43019 probability value = 2.4179148E-4

node number = 81309 probability value = 0.001411111

node number = 81657 probability value = 7.346233e-4

...

Standard English translation probability distribution two-dimensional vector:

Node number = 2200 Probability value = 2.9596334E-4

Node number = 98019 Probability value = 5.4436113E-4

Node number = 59686 Prob value = 9.3413435E-4

node number = 67669 probability value = 0.0016846128

node number = 34130 probability value = 3.4881395E-4

Node number = 91621 Probability value = 0.0010200165

node number = 75648 probability value = 3.7788178E-4

Node number = 46900 Probability value = 6.5931125E-4

Node number = 43701 Probability value = 3.1578232E-4

Node number = 101190 Probability value = 5.813566E-4

Node number = 30921 Probability value = 3.8977317E-4

node number = 45290 probability value = 0.0020875987

node number = 109167 probability value = 6.540261e-4

node number = 66046 probability value = 6.196732E-4

node number = 107567 probability value = 5.602163E-4

node number = 46880 probability value = 3.1049727e-4

Node number = 53598 Probability value = 2.5500412E-4

node number = 104364 probability value = 0.0020572094

node number = 38884 probability value = 2.3386389E-4

Node number = 10136 Probability value = 4.1619848E-4

node number = 78469 probability value = 3.9241568E-4

node number = 56445 probability value = 0.0011851747

node number = 58041 probability value = 4.1619848E-4

node number = 42070 probability value = 4.716916E-4

Node number = 13323 Probability value = 3.0124842E-4

Node number = 115514 Probability value = 2.9332083E-4

Node number = 62812 Probability value = 3.7523927E-4

Node number = 53222 Probability value = 2.4839782E-4

node number = 45236 probability value = 4.18841e-4

Node number = 24464 Probability value = 2.6293172E-4

Node number = 101119 Probability value = 0.002129879

node number = 97923 probability value = 4.0827086E-4

node number = 86740 probability value = 0.001301446

node number = 73962 probability value = 5.1265076E-4

node number = 80341 probability value = 2.8803575E-4

node number = 32429 probability value = 0.0033599767

node number = 112272 probability value = 4.0430707E-4

node number = 46792 probability value = 0.0010358717

Node number = 30814 Probability value = 3.567415E-4

node number = 61155 probability value = 0.0028195793

node number = 37190 probability value = 5.958905E-4

Node number = 91486 Probability value = 0.002388847

node number = 43574 probability value = 0.0017638888

node number = 24409 probability value = 3.1974612E-4

Node number = 61139 Probability value = 5.6946516E-4

node number = 59537 probability value = 6.196732e-4

node number = 67519 probability value = 3.54099E-4

Node number = 61129 Probability value = 2.4707653E-4

node number = 88276 probability value = 2.906783E-4

Node number = 439 Probability value = 3.4485015E-4

...

The semantic similarity of the translated English translation: 78.5 points.

Step 4: Execute the "Translation English Translation Quality Analysis Result Generation Module"

The English translation quality analysis result generation module is to synthesize the English translation fidelity analysis results output in the second step and the English translation semantic similarity analysis results output in the third step to generate and output the English translation quality score and comments. The English translation quality analysis result generation format of this embodiment is as follows:

Translated English translation quality score: 70.8 points.

Comments on the quality of the translated English translation: the English translation is consistent with the standard English translation and can correctly express the meaning of the original text.

Claims (8)

1. a kind of translator of Chinese into English English translation mass analysis method, it is characterized in that：Including one by being linked in sequence Translate English translation pretreatment module, translation English translation informativeness analysis module, the semantic similarity analysis of translation English translation The analysis model of module, translation English translation quality analysis results generation module composition, its analysis method comprise the following steps：

(1) English translation pretreatment module input translation English translation and standard English translation are translated, to translation English translation and Standard English translation is segmented respectively, word small letter, removal stop words；To participle, word small letter, remove stop words Translate English translation and standard English translation carries out part-of-speech tagging, part of speech disambiguation, phrase stripping and slicing；Export turning over for above-mentioned two-step pretreatment Translate the pre-processed results of English translation and standard English translation；

(2) it is pre- that English translation informativeness analysis module input translation English translation, standard English translation, translation English translation are translated Result, standard English translation pre-processed results, numbering translation English translation, standard English translation, translation English translation are pre- Word in result, standard English translation pre-processed results；To the word and standard English translation in translation English translation In word carry out accurate matching and the successful word of record matching and its numbering；To translation English translation and standard English translation Accurately match remaining word and carry out stem matching and the successful word of record matching and its numbering；To translation English translation and mark Quasi- English translation stem matches remaining word and carries out synonym matching and the successful word of record matching and its numbering；5th, Translated using the word matched rate accuracy rate weight and translation English of training translation English translation collection and the generation of standard English collection of translations The total words that the match is successful in text calculate and export the informativeness of translation English translation；

(3) English translation semantic similarity analysis module input translation English translation, standard English translation part of speech disambiguation knot are translated Fruit；Searched according to part of speech disambiguation result and record word corresponding node serial number in English Semantic dictionary；In English Semantic word In allusion quotation, from translation English translation word corresponding to node serial number find standard English word corresponding to node serial number, record On accessed path corresponding to each word between node serial number and node directed edge probability；According to each word on accessed path Word probability distribution bivector in English translation is translated in the probability of directed edge between corresponding node serial number and node, generation, And iterate to calculate out word probability in standard English translation and be distributed bivector；Merge word probability in translation English translation to be distributed Bivector obtains translating English translation probability distribution bivector, and is carried from translation English translation probability distribution bivector Take word probability value, generation translation English translation probability distribution one-dimensional vector；Merge standard English translation word probability distribution two Dimensional vector obtains standard English translation probability distribution bivector, and is extracted from standard English translation probability distribution bivector Word probability value, generate standard English translation probability distribution one-dimensional vector；According to translation English translation probability distribution one-dimensional vector Calculated with standard English translation probability distribution one-dimensional vector and export translation English translation semantic similarity；

(4) the output knot of English translation quality analysis results generation module input translation English translation informativeness analysis module is translated Fruit, the output result for translating English translation semantic similarity analysis module；According to translation English translation informativeness analysis module Output result, translate English translation semantic similarity analysis module output result generation translation English translation mass fraction and Comment.

2. analysis method according to claim 1, it is characterized in that：Described translation English translation pretreatment module processing step It is rapid as follows：

P201 starts；

P202 reads in translation English translation and standard English translation；

P203 judges to translate English translation and whether standard English translation quantity is identical, if it is turns P204, otherwise turns P202；

P204 is to translation English translation participle and will translate the word small letter after English translation segments；

The word segmentation result of P205 output translation English translations and the word small letter result of translation English translation；

P206 statistical translation English translation total words；

P207 carries out part-of-speech tagging to the word segmentation result for translating English translation；

P208 carries out phrase stripping and slicing to the part-of-speech tagging result for translating English translation；

P209 carries out part of speech disambiguation to the phrase stripping and slicing result for translating English translation；

The part-of-speech tagging result of P210 output translation English translations, the part of speech disambiguation result for translating English translation, translation English are translated The phrase stripping and slicing result of text；

P211 to standard English translation segment and by standard English translation segment after word small letter；

The word segmentation result of P212 outputting standard English translations and the word small letter result of standard English translation；

P213 SSs English batch translation total words；

P214 carries out part-of-speech tagging to the word segmentation result of standard English translation；

P215 carries out phrase stripping and slicing to the part-of-speech tagging result of standard English translation；

Phrase stripping and slicing results of the P216 to standard English translation carries out part of speech disambiguation；

The part-of-speech tagging result of P217 outputting standard English translations, the part of speech disambiguation result of standard English translation, standard English are translated The phrase stripping and slicing result of text；

P218 terminates.

3. analysis method according to claim 1, it is characterized in that：Described translation English translation informativeness analysis module Calculation formula is defined as follows：

(1) word matched accuracy rate calculation formula

Word matched accuracy rate refers to translate word matched success number and translation English between English translation and standard English translation The ratio of total words in language translation, its calculation formula are as follows：

(2) word matched recall rate calculation formula

Word matched recall rate refers to translate word matched success number and standard English between English translation and standard English translation The ratio of total words in language translation, its calculation formula are as follows：

(3) harmonic-mean calculation formula

Harmonic-mean refer to the word matched accuracy rate of calculation formula (1) and calculation formula (2) word matched recall rate it Between average value, its calculation formula is as follows：

In calculation formula (3), word matched success rate is calculated by calculation formula (1), and word matched recall rate is by calculating Formula (2) is calculated, and a is the weight of word matched accuracy rate, and 1-a is the weight of word matched recall rate.

(4) penalty coefficient value calculation formula

Penalty coefficient value is to reduce to miss between translation English translation content and standard English translation content caused by harmonic-mean The coefficient of difference, its calculation formula are as follows：

In calculation formula (4), phrase number refers to translate the sum of noun phrase and verb phrase in English translation；B is punishment The adjusting parameter of coefficient value size；C is phrase number and the adjustment of the word number ratio size that the match is successful in translation English translation Parameter；B and c is calculated by translation English translation informativeness analysis module.

(5) English translation informativeness calculation formula is translated

Translation English translation informativeness refers to semanteme of word similarity degree of the word with standard English translation for translating English translation, Its calculation formula is as follows：

Translate English translation informativeness=(1- penalty coefficients value) × harmonic-mean (5)

In calculation formula (5), harmonic average numerical value is calculated by calculation formula (3), and penalty coefficient value is by calculation formula (4) It is calculated.

4. analysis method according to claim 1, it is characterized in that：At described translation English translation informativeness analysis module It is as follows to manage step：

P301 starts；

P302 reads in the word segmentation result of translation English translation and the word small letter result of translation English translation；

P303 reads in the word segmentation result of standard English translation and the word small letter result of standard English translation；

P304 is according to the word segmentation result of translation English translation and word small letter result, the standard English translation of translation English translation Word segmentation result and standard English translation word small letter result, generation translation English translation-standard English translation text It is right；

P305 reads in the text pair of one group of translation English translation-standard English translation；

P306 starts from scratch the text pair translation English translation each word for translating English translation-standard English translation volume Number；

P307 starts from scratch the text pair standard English translation each word for translating English translation-standard English translation volume Number；

The text pair for translating English translation-standard English translation is translated each word of English translation, and translation English by P308 The each word of standard English translation of the text pair of language translation-standard English translation is accurately matched；

The text pair translation English translation of P309 statistical translations English translation-standard English translation, and translation English translation- The accurate word that the match is successful of the text pair standard English translation of standard English translation and numbering；

P310 judges whether the text pair translation English translation for translating English translation-standard English translation also has inaccurate matching Word, if it is turn P308, otherwise turn P311；

The text pair for translating English translation-standard English translation is translated English translation, and translation English translation-mark by P311 The text pair standard English translation of quasi- English translation carries out stemmed without the accurate word that the match is successful；

Word in the text pair translation English translation of translation English translation-standard English translation after P312 will be stemmed, and The word of the text pair standard English translation of translation English translation-standard English translation after stemmed carries out stem matching；

The text pair translation English translation of P313 statistical translations English translation-standard English translation, and translation English translation- The text pair standard English translation stem word that the match is successful and numbering of standard English translation；

P314 judges whether the text pair for translating English translation-standard English translation is translated also not stemmed in English translation The word of matching, if it is turn P312, otherwise turn P315；

The text pair for translating English translation-standard English translation is translated English translation, and translation English translation-mark by P315 The text pair standard English translation of quasi- English translation does not have the stem word that the match is successful to carry out meaning of a word lexical or textual analysis；

P316 translates the text pair that English translation-standard English translation is translated after the meaning of a word lexical or textual analysis word of English translation, And after meaning of a word lexical or textual analysis translate English translation-standard English translation text pair standard English translation word carry out it is synonymous Word matches；

The text pair translation English translation of P317 statistical translations English translation-standard English translation, and translation English translation- The text pair standard English translation synonym word that the match is successful and numbering of standard English translation；

P318 judges that the text pair for translating English translation-standard English translation translates in English translation whether also have non-synonym The word of matching, if it is turn P315, otherwise turn P319；

P319 judges whether the text pair for the translation English translation-standard English translation do not corrected also, if it is turns P305, Otherwise P320 is turned；

P320 reads in the text pair translation English translation word of translation English translation-standard English translation, and translation English The result that the text pair standard English translation word of translation-standard English translation accurately matches；

P321 reads in the text pair translation English translation word of translation English translation-standard English translation, and translation English The result of the text pair standard English translation word stem matching of translation-standard English translation；

P322 reads in the text pair translation English translation word of translation English translation-standard English translation, and translation English The result of the text pair standard English translation word synonym matching of translation-standard English translation；

P323 marks the text pair translation English translation word that the match is successful of translation English translation-standard English translation；

P324 reads in the text pair translation English translation total words of translation English translation-standard English translation；

P325 reads in the text pair standard English translation total words of translation English translation-standard English translation；

P326 translates the text pair for translating English translation-standard English translation the total words of English translation, and translation Total words in the text pair standard English translation of English translation-standard English translation, translation English translation-standard English are translated The text pair translation English translation of text accurately matches, stem matches, the synonym total words that the match is successful, substitutes into translation English Calculation formula (1), the calculation formula (2) of language translation informativeness analysis module calculate translation English translation-standard English translation Text pair translates the accuracy rate and recall rate of English translation word matched；

P327 is using translating English translation collection and the training of standard English collection of translations draws optimized parameter a, b, c value；

The text pair of a value, translation English translation-standard English translation is translated the accurate of English translation word matched by P328 Rate, the recall rate for the text pair translation English translation word matched for translating English translation-standard English translation substitute into translation English The calculation formula (3) of language translation informativeness analysis module calculates harmonic-mean；

The value of parameter b, c, the text pair of translation English translation-standard English translation are translated the number of English translation phrase by P329 Amount, translation English translation-standard English translation text pair translation English translation accurately match, stem matching, synonym With successful total words, the calculation formula (4) for substituting into translation English translation informativeness analysis module calculates penalty coefficient；

By harmonic-mean, the calculation formula (5) that penalty coefficient substitutes into translation English translation informativeness analysis module calculates P330 Translate English translation informativeness；

P331 terminates.

5. analysis method according to claim 1, it is characterized in that：Described translation English translation semantic similarity analysis mould The calculation formula of block is defined as follows：

(1) ProbabilityDistribution Vector calculation formula

ProbabilityDistribution Vector refers in English Semantic dictionary, from node checks standard English corresponding to translation English translation word Node corresponding to word, on accessed path corresponding to each word between node serial number and node the probability of directed edge form to Amount, its calculation formula are as follows：

ProbabilityDistribution Vector_t=(1-a) × adjacency matrix × ProbabilityDistribution Vector_t-1+ a × ProbabilityDistribution Vector₀ (6)

In calculation formula (6), t represents the number searched, and a represents the probability of directed edge on the t times accessed path, adjacency matrix Represent the word adjacency matrix in English Semantic dictionary, ProbabilityDistribution Vector₀Represent the probability distribution of lookup original position node Vector, ProbabilityDistribution Vector_t-1Represent the ProbabilityDistribution Vector of the t-1 times lookup node, ProbabilityDistribution Vector_tRepresent to look into for the t times Look for the ProbabilityDistribution Vector of node.

(2) English translation Semantic Similarity Measurement formula is translated

Translation English translation semantic similarity refers to translate semantic similar between English translation content and standard English translation content Degree, its calculation formula are as follows：

In calculation formula (7), translation English translation probability distribution one-dimensional vector refers in English Semantic dictionary, from translation English Node serial number corresponding to language translation word searches node serial number corresponding to standard English word, and each word is corresponding on accessed path Node between directed edge probability form one-dimensional vector；Standard English translation probability distribution one-dimensional vector refers in English language In adopted dictionary, node serial number corresponding to standard English word is searched from node serial number corresponding to translation English translation word, is searched The one-dimensional vector that the probability of the directed edge of path end position node connection is formed；| | translation English translation probability distribution it is one-dimensional to Amount | | refer to the mould for translating English translation probability distribution one-dimensional vector；| | standard English translation probability distribution one-dimensional vector | | refer to The mould of standard English translation probability distribution one-dimensional vector.

6. analysis method according to claim 1, it is characterized in that：Described translation English translation semantic similarity analysis mould Block processing step is as follows：

P401 starts；

P402 reads translation English translation word part of speech disambiguation result；

P403 ergodic translation English translation word part of speech disambiguation results；

P404 according to translation English translation word part of speech disambiguation result search and record word in English Semantic dictionary corresponding to Node serial number；

P405 judges whether the translation English translation word part of speech disambiguation result not traveled through also, if it is turns P403, otherwise Turn P406；

P406 reads standard English translation word part of speech disambiguation result；

P407 travels through standard English translation word part of speech disambiguation result；

P408 according to standard English translation word part of speech disambiguation result search and record word in English Semantic dictionary corresponding to Node serial number；

P409 judges whether the standard English translation word part of speech disambiguation result not traveled through also, if it is turns P407, otherwise Turn P410；

It is corresponding that P410 searches standard English word in English Semantic dictionary, from node serial number corresponding to translation English translation word Node serial number；

P411 records on accessed path the probability of directed edge between node serial number and node corresponding to each word；

The probability of P412 directed edges between node serial number and node according to corresponding to each word on accessed path, generation translation English Language translation word probability is distributed bivector；

P413 calculates single in standard English translation according to the calculation formula (6) of translation English translation semantic similarity analysis module Word probability is distributed bivector；

P414 judges to translate in English translation whether also have the word do not searched, if it is turns P410, otherwise turns P415；

P415 merges translation English translation word probability distribution bivector and obtains translating English translation probability distribution bivector；

P416 extracts word probability value, generation translation English translation probability point from translation English translation probability distribution bivector Cloth one-dimensional vector；

P417 merges standard English translation word probability distribution bivector and obtains standard English translation probability distribution bivector；

P418 extracts word probability value, generation standard English translation probability point from standard English translation probability distribution bivector Cloth one-dimensional vector；

P419 is calculated according to the calculation formula (7) of translation English translation semantic similarity analysis module and is exported translation English translation Semantic similarity；

P420 terminates.

7. analysis method according to claim 1, it is characterized in that：Described translation English translation quality analysis results generation The calculation formula of module is defined as follows：

(1) English translation mass fraction calculation formula is translated

Translation English translation mass fraction refers to the content journey similar to the contents semantic of standard English translation for translating English translation Degree, its calculation formula are as follows：

English translation mass fraction=0.5 × English translation+0.5 × English translation of informativeness semantic similarity (8)

In calculation formula (8), translation English translation informativeness score is calculated by calculation formula (5), translates English translation Semantic similarity is calculated by calculation formula (7).

8. analysis method according to claim 1, it is characterized in that：Described translation English translation quality analysis results generation Resume module step is as follows：

P501 starts；

P502 reads translation English translation informativeness；

P503 reads translation English translation semantic similarity；

P504 is calculated according to the calculation formula (8) of translation English translation quality analysis results generation module and is exported translation English and translates Literary mass fraction；

P505 is according to translation English translation mass fraction output translation English translation comment；

P506 terminates.