TW200521712A - System, method and machine-readable storage medium for automated sentence annotation - Google Patents

Abstract

A system and a method for automated sentence annotation comprise a sentence annotation module for receiving a sentence, in which, according to a finite state machine containing a plurality of orderly states, where there is a conversion word between two states, conversion words in the sentence is detected in order. If there are conversion words in the sentence containing same order, the words between the two conversion words in the sentence are annotated into corresponsive states and thus structural information containing states and words between the two conversion words in the sentence is generated.

Description

200521712 V. Description of the invention (1) The technical field to which the invention belongs This invention is a system and method for automatic tagging of documents, in particular a system and method for automatically tagging sentences using a finite state machine. u In the prior art, due to the polysemy of words and the polysemy of words, traditional information retrieval (inf 〇rmati 〇nretrieva 1) technology encountered this bottleneck. Therefore, many automatic file tagging technologies have been proposed, adding one file to one file. The semantic mark of the communist drought is used to improve precision (reca 1 1) and cross-domain extensibility (sca 1 ibi 1 ity). Traditional file automatic tagging technology can be divided into three types: statistics — based, machine learning, and heur i st ic rule. The basic statistical technology uses a large number of corpora (thesaurus) to analyze statistical words to retrieve information, but its accuracy will be affected by the size of the corpus. Machine learning technology proposes a mathematical model for the machine to learn automatically to identify information, which requires a certain training time ', but sometimes the training results will not converge, which will lead to unstable accuracy. The exploration rule technology is simple and easy to understand, which is more in line with human understanding, but requires a lot of manual manual fine-tuning, otherwise the accuracy rate will not be easily improved. Although ‘traditional document auto-tagging technology’ can moderate the problem of linguistic polysemy and polysemy. However, because it mostly marks the entire document (not a sentence), its accuracy improvement in information retrieval can only stay at the document level. In addition, syllables, homonyms, and wrong phrases sometimes appear in sentences.

0213-A40162TWF (Nl); B9250TW; SNOWBALL.ptd Page 5 200521712 V. Description of the invention (2) Combined with the situation, the accuracy of sentence marking is reduced. Therefore, a system and method are needed to improve the accuracy of information retrieval in the document. In addition, the accuracy of the = mark with a combination of words, homonyms, and wrong words is required. In view of the above, the content of the low and medium sentence of Bu Yi I is to provide a kind of sentence / system and method to improve the accuracy of information retrieval. In addition, there are different sounds, syllables, and homonyms in :: Words and wrong word phrases; Mouth: Does not reduce the accuracy of sentence tagging. First, the sentence automatic tagging system and method of Ϊ, enter the device, and use a bus to link it in one # 存 U: 输 一 # material warehouse system, file system, or other storable ^ σ to include A domain thesaurus, with 1 'temple, among many synonyms, syllables and homonyms. # 库 中 录 一词 The software module architecture includes a program script group, a synonym processing module, and a homonym;; == a memory module and a misaligned word processing module. ,,,,, sentence tag module included in the Japanese word processing body, the synonym processor to load the memory management module, the different word processing module, and calcium = (and, the same word according to the program instructions And the user uses the input sink processing module, parallel sentence automatic tagging function, and the data input under the best circumstances, and displays it on the display device. / 卜 ’will be the result after execution

0213 -A40162TWF (N1); B9250TW; SNOWBALL.ptd Page 6 200521712 V. Description of the invention (3) The finite state machine includes several states, including a specific ending state, which represents the semantic marking function of the entire sentence End. In a finite state machine, a transition word is required to enter from one state to another. The synonym processing module is used to input a word, search the domain lexicon, and output synonyms related to the word. Synonyms refer to different words with the same meaning. The homophone processing module is used to input a word, search the domain lexicon, and output the homophone related to the word, and the heterophone processing module is used to search the domain vocabulary, and output the different pronunciation related to the word. Words, in which homophones and heterophones represent misspellings in sentences that may occur due to incorrect input of a single word. The misplaced word processing module is used to input a word, recombine the words contained in the word, and generate and output a misplaced word. The misplaced word represents a mistake that may occur in the sentence due to the reversal of the word order word. The sentence mark module is used to input the finite state machine, and the conversion words in it are sequentially input to the synonym processing module, homophone processing module, heterophone processing module, and misaligned word processing module to obtain each conversion. Synonyms, homophones, heterophones, and misaligned words of words form a set of synonyms and a set of misspellings. Enter a sentence, in order ^ [converted words, synonyms, homophones, homonyms, or misaligned words that exist in the Zhentest sentence, mark the corresponding states between the two converted words, and save them as Extensible Markup Language (XML) messages, corrupt database records, or audit records. Embodiment 1 FIG. 1 shows an automatic sentence marking system according to an embodiment of the present invention.

0213-Α40162TWF (N1); B9250TW; SNOWBALL.p t d p. 7 200521712 V. System architecture diagram of invention description (4). An automatic sentence marking system according to an embodiment of the present invention] includes a storage device 11, a central processing unit 2, a memory 1, 3, a display 15, an input device 15, and is connected together using a bus 16 . Combining the storage device 11, the central processing unit 丨 2, the memory 丨 3, the display device i 4 and the input device 15 can form a mainframe, a personal computer, a workstation, a notebook computer or other computer equipment.

The storage device 11 may be a database system, a file system, or another device capable of storing lean materials, which includes a domain thesaurus (domai η thesaurus), and a plurality of synonyms, syllables and other words recorded in the thesaurus. And homonyms. For example, the word "Xing Jing" in the domain thesaurus contains two synonymous lines "to" and "quotation"; two homonyms "type Jing", "quoting" and "different words"; π 袭 经 &#; FIG. 2 is a diagram showing a software module structure of an automatic sentence marking system according to an embodiment of the present invention. The software module structure of the embodiment of the present invention exists in Lu Jiyi's body 13 Sentence mark module 1 3 composed of instruction code 1. Synonym processing module 3 2. Homophone processing module 3 3. Homophone processing module 1 3 4 and wrongly arranged word processing module 3 5. Central processing

Is 1 2 is used to load the sentence mark module contained in memory 1 3 丨 3 1. Synonymous processing module 1 3 2. Homonym processing module 丨 3 3

Group 1 34 arranges word processing modules 4 and 4 with errors 4 and 35, and executes the automatic sentence marking function according to the program instructions and the data entered by the input device 15, and in the best case, will be executed after The results are displayed on the display device 14. Figure 3 shows an exemplary finite state machine diagram according to an embodiment of the present invention.

200521712 V. Description of Invention (5) Intent. The finite state machine 3 includes five states, namely a person 321, an incident time 322, an f-issue model 323, an incident place 324, and 331. Among them, the end state 331 is a special state, which represents the end of the entire sentence. In the finite state machine, from one state to another, the need-transformation word is, in this embodiment, "quotee" 311, "Yu" 312, ', driving car number "313 ,,, Passing through 314 and Investigating " 315. The finite state machine 3 can be expressed as follows: {subject to punishment + {at} + incident time + {driving car number 丨 + incident model + {line association + {by police check} + end). 、、 二} Case ♦ Place The synonym processing module 132 is used to input a word, and to check out the synonymous words that are related to the word. Synonyms and fingers have the same meaning. The homophone processing module 丨 33 is used to input a word, and to output the homophone related to the word, and the heterophone processing module is used to output the heterophone and complex related to the word. Then the syllables and syllables are representative of the possible misspellings in a sentence: the second word appears in the second sentence. Wrongly arranged word processing module 丨 3, a decisive word, recombining the words contained in the word to generate and # ^ 入 word, where the wrongly arranged words represent = wrong in the sentence A typo that appears. This is because the early word order reverses sentence marking module 1 3 1 to input the finite state machine 3, and the words 311, 312, 313, 314, and 315 are used to input synonym ^, household management group 132, and homonyms in order. The processing module 133, the different sound word processing mode 2 and the wrongly arranged word processing module 丨 35, get each conversion gas, /, word, homophone, different phonetic word, and wrongly arranged word, which form 3% —same Synonym

200521712 V. Description of the invention (6) Collection and a typo collection. FIG. 4 is a schematic diagram showing a set of example synonyms and misspellings according to an embodiment of the present invention, and the conversion words 3 π, 3 丨 2, 3 1 3, 3 1 4 and 3 1 5 are synonyms, homophones, and heterophones. And the misplaced word 'is listed in columns 471, 472, 47 3, 474, and 475. It must be noted here that any conversion word does not necessarily need to be related to synonyms, homonyms, homonyms, and misaligned words. It can generate related synonyms, homonyms, homonyms, or errors through user settings. The permutation group 131 enters the drunk driving judgment sentence as shown below. Λ ΛΛΛ 90th Anniversary Road-Section Wusan Lane Alley 'Jing Xin Er; Xiao F Model' via Taipei Mayor "Taste of alcohol" is checked with breath alcohol tester; 丨 [Currently, the person who carries the disposition has a degree of 0.5 mg per liter. The person who breathes the person's breath contains a conversion word of the Chinese character 3, the related affiliation ,. Next, according to the limited state record, mark the names of the person 3 2 2 and the misspellings in the sentence, and carry out the sentence mark 323, the incident location 324, etc. 4 状, the incident time 322, the incident model sentence marking module process details As described below. The word &quot; subjected person &quot;, therefore, in &quot; is the second text, one of the conversion words begins to exist, in ', 312, and in the sentence after detecting ^ two π, look for the following In the sentence, look for the conversion word &quot; in the two texts &quot; in π, which does not exist in "于" &quot; driving car number &quot; Yiqi, and car number 3 1 3. Because of any synonyms in the sentence, in Γ in :, and the conversion word 313 does not exist in the related misspell &quot; drive number, or &quot;丨 丨 words. As a result, the detection 0213-Α40162TWF (N1); B9250TW; SNOWBALL.ptd Page 10 200521712 V. Description of the invention (7) = which: contains &quot; driving number ,, the word 'after the gang, detect the conversion word || "Scripture" 314, driving bluffing &quot; 315. As a result, in the following sentence, the Jing Jing police switched over and found that the homophone type script &quot; and the conversion word &quot; 经 ^ 到 = 与 词 二 经 &quot; 314 It was found by police interception. "When it was found, it was" synonymous with M-3131 5 ". The master, Ren λ α ss found out 丨 丨 3 1 5 transition 啕 Japanese representative entered the end state 331, no longer Continue the sentence bar ^ change ^, the state 2 ;;; group 131 will be the words between the two conversion words :, mark out the phase of the library phase ΐ: ;;, "ί database records, slot case records. Another Xibu , During A, you can also use punctuation or words such as stopwords that are too broad, and in the best case. <Judgment text>, the description of the XML language is as follows, <Person name> Wu Adai &lt; / person name &gt; <incident time> 90 years of the Republic of China +, Xu &lt; / incident time> 90 0:33 am on October 13th, 2015. <f-type vehicle> LF c ——No. 6 passenger car &lt; / incident model &gt; <occurrence location> Wusan Lane, Section 1, Chang'an East Road, Taipei City &amp;口 〇Place of occurrence &lt; / decision text>

It must be noted here that if the marked sentence cannot enter the end state

200521712 V. Description of the invention (8) ----- Method flow chart, this method consists of code, and can be loaded and executed by the central processor. First, as in step S5 1 1, a finite state machine is received. As shown in FIG. 3, the finite state machine 3 includes five states, which are a person's name 3 2 1, an incident time 3 2 2, and an incident. Model 3 2 3, place of crime 3 2 4 and end 3 3 1. Among them, the end state 3 31 is a special state, which represents the end of the semantic marking function of the entire sentence. In a finite state machine, transitioning from one state to another requires a transition word. "▲ As shown in step S521 and step S522, the synonyms and misspells corresponding to the converted words are obtained, where the misspelled words include homophones, insyllabics, and wrongly arranged words. The results are shown in FIG. 4 As shown in step S531, a sentence is received, and then the conversion word in the sentence is detected as shown in step S541. The detection method can compare the conversion word itself, a synonym or a typo corresponding to the conversion word. In this case, in this step, the conversion word itself is compared first, followed by the synonyms, and finally the misspelled word. As shown in step S542, 'judgment whether the sentence ends, if yes, then the entire method is terminated; otherwise, step S543 is performed. As step S543 As shown in the figure, if it is judged that the end state is 'Yes', then step S551 is performed; otherwise, it returns to step S541 and continues to detect the next converted word. As shown in step S551, mark the sentence between the converted word and the converted word appropriately. Status, such as the name of the person 3 2], the time of the incident 3 2 2, the model of the incident 3 2 3, the place of the incident 3 2 4 and so on, and a structured message, a database record (rec 0rd), or a file Record. In the preferred case, the structure of this information

200521712 V. Description of the invention (9) The information is an XML message. Furthermore, the present invention proposes a computer-storable-computer program 'the above-mentioned computer program is used to implement the storage method'. This method will perform the steps described above. . Fig. 6 is a schematic diagram showing a computer-readable storage medium according to an embodiment of the present invention. This automatic memorization method is a computer program 620, which is used to implement sentence-storage media 60. The storage format contains six logics, namely, the receiver automatically = 625 and the structured data logic 626. The conversion word logic of σ is therefore based on the automatic sentence analysis of the sentence provided by the present invention: in order to improve the accuracy of information retrieval, in addition to ": and side ;; =, homophones and wrong word combinations; Sentence: Intermediate output reduces the accuracy of sentence markings. "Brother, it will not be disclosed though the embodiment of the present invention is as above, but it is not an invention." Anyone who is familiar with this skill will not depart from the scope of the present invention. When some changes and retouching can be done, therefore, the spirit and scope of the present invention are determined by the scope of the attached patent application. ', 濩 Scope when 0213-A40162TWF (N1); B9250TW; SNOWBALL.

Ptd Page 13 200521712 Schematic description In order to make the above-mentioned objects, features and advantages of the present invention more comprehensible, the following specific examples are given in conjunction with the accompanying drawings for detailed description as follows: Figure 1 shows the basis System architecture diagram of a sentence automatic tagging system according to an embodiment of the present invention; FIG. 2 is a diagram showing a software module architecture diagram of a sentence auto-tagging system according to an embodiment of the present invention; and FIG. 3 is a diagram showing an exemplary limited state according to an embodiment of the present invention Machine diagram; Figure 4 is a schematic diagram showing a set of example synonyms and misspellings according to an embodiment of the present invention; Figure 5 is a flowchart showing a method for automatically marking a sentence according to an embodiment of the present invention; Schematic diagram of computer-readable storage medium of automatic sentence marking method in the embodiment of the invention. Explanation of symbols 1 0 ~ sentence automatic tagging system; 11 ~ storage device; 1 2 ~ central processing unit; 1 3 ~ memory; 1 4 ~ display device; 1 5 ~ input device; 1 6 ~ busbar; 1 3 1 ~ Sentence Marker Module; 1 3 2 ~ Synonym Processing Module;

0213-A40162TWF (Nl); B9250TW; SNOWBALL.ptd Page 14 200521712 Schematic brief description 1 3 3 ~ homophone processing module; 1 3 4 ~ heterophone processing module; 1 3 5 ~ misaligned word processing Module; 311 &gt; 312 ..... 3 1 5 ～ converted words; 3 2 1 ～ person name status; 3 2 2 ～ incident time status; 3 2 3 ～ incident model status; 3 2 4 ～ incident Location status; 3 3 1 ~ End status; 4 1 ~ Conversion word field; 4 2 ~ Synonym field; 4 3 ~ Sentence word field; 4 4 ~ Homophone word field; 4 5 ~ Hardphone word field Bits; 4 6 ~ Wrongly arranged word fields; 471, 472 ..... 475 ~ Synonyms of conversion words and misspellings S51 1, S521 ..... S55 Operation steps; 6 2 0 ~ Sentences automatically Mark computer program; 6 2 1 ~ Receive finite state machine logic; 6 2 2 ~ Get synonym logic related to converted words; 6 2 3 ~ Get wrong word logic related to converted words; 624 ~ Receive sentence logic; 6 2 5 ~ Detect conversion logic in sentences; 6 2 6 ~ Generate structured data logic.

0213-A40162TWF (N1); B9250TW; SNOWBALL.ptd Page 15

Claims (1)

200521712 6. Scope of patent application1. An automatic sentence marking system, including: a sentence marking module for receiving a sentence, based on a finite state machine containing a plurality of sequential states, wherein a transition is included between the two states Word 1 sequentially detects the corresponding conversion words in the above sentence. If the above sentence contains all the conversion words in the same order, the words between the corresponding two conversion words in the sentence are marked as the corresponding status and Generate structured data containing one of the above-mentioned states and one of the words between the two above-mentioned converted words in the above-mentioned sentences. 2 · The sentence automatic tagging system as described in item 1 of the scope of patent application, wherein the structured data is a database record, a file record, or an extended mark (X red) message. 3. Automatic sentence marking system as described in item 1 of the scope of patent application, word. &lt; Synonym 4 of 锝 change 3 · As in the third paragraph of the scope of the patent application, the sentence is automatically marked in the sentence mark module, and the words corresponding to the above mentioned conversion words &amp; Detect the corresponding conversion word in the above sentence and the above-mentioned synonym of the conversion word in the same sentence in order, if the above sentence contains the above conversion word that should be in the previous order and the above-mentioned [same sequence, Then one of the two corresponding conversion words in the above sentence ^ = should be in the above state. H is recorded as phase 5. The sentence described in item 1 of the scope of patent application automatically includes a storage device for storing a plurality of words corresponding to the above system. Mistakes in Runes 0213 -A40162TWF (N1); B9250TW; SNOWBALL.ptd page 16 200521712 Patent application scope 6 · Sentence automatic tagging system as described in item 5 of the patent application scope, where the above misspellings are homonyms and a disyllabic words Or a misaligned word. 7) According to the automatic sentence marking system described in Item 5 of the scope of patent application, in the sentence marking module, search for the above misspellings corresponding to the above conversion words, and sequentially detect the corresponding above conversion words in the above sentences and the corresponding In the above misspellings of the above conversion word, if the above sentence contains one of all the above conversion words in the same order and the above error words corresponding to the above conversion word, then the corresponding two above conversion words in the above sentence are separated. The word is marked as above. 8 · According to the automatic sentence marking system described in item 4 of the scope of patent application, the storage device further stores a plurality of misspellings corresponding to the above conversion words. 9. According to the automatic sentence marking system described in item 8 of the scope of patent application, in the sentence marking module, the above-mentioned synonyms corresponding to the above-mentioned conversion words and the above-mentioned misspellings' are sequentially detected in the above sentences. Words, the above synonyms corresponding to the above conversion words, and the above typos corresponding to the above conversion words, if the sentence contains all the above conversion words in the same order, the above synonyms corresponding to the above conversion words, and the above corresponding to the above conversion words. One of the misspellings marks the word between the corresponding two converted words in the above sentence as the above state. I 0 · The sentence automatic tagging system as described in item 9 of the scope of the patent application, wherein the misspelling is a homonym, a / unison, or a wrong arrangement. II · A method for automatic tagging of sentences, being-with a central processing unit 0213-A40162TWF (Nl); B9250TW; SNOWBALL.ptd Article 17 Yu 200521712 VI. Implementation of a patent-applied electronic device, the method includes the following steps: · Receive a sentence; According to a finite state machine containing a plurality of sequential states, A conversion word is included between the two states, and the corresponding conversion words in the sentence are sequentially detected; and if the sentence contains all the conversion words in the same order, the corresponding two conversion words in the sentence are replaced. The inter-words are marked as corresponding to the above-mentioned states and a structured data including one of the above-mentioned states and the word between the two above-mentioned converted words in the sentence is generated. 12 · The automatic tagging method for sentences as described in item 11 of the scope of patent application, wherein the structured data is a database record, a file record, or an extensible markup language (XML) message. Go to 1 to 3 parties as described in # · 申 Λ Patent Range Item No. 11's method for automatic tagging of sentences. The method further includes the following steps: Retrieve a synonym corresponding to one of the above conversion words. • Detect the above sentence in sequence. 'The above-mentioned synonyms of the conversion word; the == conversion word and the above-mentioned conversion word corresponding to the order of j in the above-mentioned synonyms with the corresponding above-mentioned conversion word in the above sentence and the corresponding word between the two above-mentioned conversion words If you mark ^^^, the method in the above sentence further includes the following steps: Retrieve the corresponding above conversion words — sequentially detect the corresponding ones in the above sentences :: other words; addictive conversion words and methods 14. If you apply The above-mentioned state corresponding to the item 11 of the patent scope. U I, —, The sentence of the auto-tagging party corresponds to the above page 18 0213-A40162TWF (Nl); B9250TW; SNOWBALL.ptd 200521712 6. The above typos of the conversion word in the scope of patent application; and if the above sentence contains All of the above-mentioned conversion words having the same order and one of the above-mentioned misspellings corresponding to the above-mentioned conversion words will mark the words between the corresponding two above-mentioned conversion words in the above-mentioned sentences as the above state. 1 5. The automatic tagging method of a sentence as described in item 14 of the scope of patent application, wherein the misspelling is a homonym, a disyllabic word, or a wrong permutation word. 1 6. The automatic tagging method for a sentence as described in item 13 of the scope of patent application, the method further includes the following steps: searching for the above synonyms of the corresponding conversion word and the above misspelling; sequentially detecting the corresponding above in the above sentence The conversion word, the above synonyms corresponding to the above conversion word, and the above typo corresponding to the above conversion word; and if the sentence contains all the above conversion words in the same order, the above synonyms corresponding to the above conversion word, and the corresponding above conversion word One of the above-mentioned misspellings marks the word between the corresponding two conversion words in the sentence as the above-mentioned state. 1 7. —A computer-readable storage medium for storing a computer program for loading into a computer system and causing the computer system to execute any one of items 11 to 16 of the scope of patent application The method described 0 0213 -A40162TWF (N1); B9250TW; SNOWBALL.p t d p.19