TWI227417B - Digital resource recommendation system, method and machine-readable medium using semantic comparison of query sentence - Google Patents

Abstract

The present invention is related to a digital resource recommendation system and a method using semantic comparison of query sentence. The invented system includes a storage device and an analysis module. The storage device stores an ontology and a plurality of historical records of query sentence. The analysis module is used to enter a new query sentence and look up a historical query sentence from the historical records of query sentence. A first vocabulary is produced from the new query sentence. A second vocabulary is produced from the historical query sentence. The ontology is used to search for a first semantic mark corresponding to the first vocabulary, and a second semantic mark corresponding to the second vocabulary, in order to calculate a semantic distance of the ontology corresponding to the first semantic mark and the second semantic mark. Subsequently, the digital resource ID codes are displayed in an ascending order according to the degree of similarity based on the semantic distances.

Description

1227417 Sun and Moon ⑴ "" ------ Technical Field to which the Invention belongs This invention of Shi Wuli is a recommendation system and method, especially an electronic resource recommendation system and method for query direction = thinking comparison. In the prior art, in order to allow users to obtain the required electronic resources, including web pages, network services, documents, etc., application systems usually aggregate these resources into one person; After that, users can obtain the appropriate resources based on this directory structure. In addition to consolidating electronic resources into a directory structure, the application system can provide a search engine (for example, tornado, Google or sweet potato vine, 1 according to the keywords or phrases entered by the user, using a search to perform a nose The law compares the content of electronic resources (for example, the content of web pages or files) with the description of electronic resources (for example, the description of Internet services), and lists similar electronic resources that produce sea bass for users to choose. ^ Recently, more and more More and more application systems will provide a recommendation analysis engine, according to the user's background (access log), to perform electricity and source recommendation, for example, the Amazon Internet Bookstore (Amaz〇n · comm) Anyway, in addition, the recommendation analysis engine can also cooperate with the search engine to search the query keywords or words entered by the user and the results, and perform electronic resource recommendation analysis. For the analysis method of electronic resource recommendation, Tongwei will compare the words and phrases newly entered by the user with the keywords to compare the words and phrases collected in the past to find out Like words, # allocated over the selection of the recommended results. Although this is a valid one recommended method

1227417 V. Description of the invention (2) The query sentence is usually quite short. Therefore, it is difficult to overcome the situation that the polysemy and the difference in the length of the sentence cause poor accuracy. Therefore, a system and method are needed to resolve the shortcomings of keyword comparison. With the accuracy of 4 plus the comparison, it is more effective to use word and sentence comparison for electronic resource recommendation. SUMMARY OF THE INVENTION In view of this, an object of the present invention is to provide an electronic resource recommendation system and method for query sentence semantic ratio, which can solve semantic problems that cannot be overcome by traditional keyword comparison technology through semantic tagging and comparison techniques. The problem of word length difference. According to the above purpose, the present invention recommends a system and method for electronic data comparison of query sentences semantically. First, a storage device, a central processing unit, a memory, a display device, and an input device are provided, and a bus is used to connect them. , 々 re-emergence. The storage device stores an ontology and a query history record. The memory contains an analysis module containing program instructions and a feedback module. The central processing unit is used to load the analysis module and the feedback module contained in the memory, and execute the query sentence semantic comparison and electronic resource recommendation function according to the program instructions and the data input by the user through the input device, and execute The subsequent results are displayed on a display device. ^ Ontology library is a conceptual architecture describing the relationship between things. There are two stops for query and sentence history, query sentence and work flow compilation. Each record stores the query statements entered by the user in the past and the workflow selected and executed by the user. The query block is used to store the query entered by the user in the past, and the workflow number field stores the selected

1227417

And execute the workflow. The electronic resource recommendation of the present invention is divided into two stages, analysis and result feedback, and the point = is implemented through the analysis module and the feedback module.

In the analysis section, the analysis module first receives the query sentence input by the user through the input device. Then perform Chinese word segmentation on the newly entered query sentence and the historical query sentence in the historical record of the query sentence, and the words containing time in the sentence are ignored, and the "subject" — verb (verb)-subject (Ob ject) " sentence pattern sequence. Next, from the parsed " subject-verb-acceptor " Departments, documents, and concepts in the concept tree, attributes, and entities. Analysis module according to the semantic labeling of the newly input query sentence, comparison

In the query history records, the semantic meaning of all records is calculated, and the semantic distance is calculated. Semantic distance calculation is divided into two stages, conceptual distance calculation and physical distance calculation. The concept distance calculation includes (1) the semantic distance between subject concepts, (2) the semantic distance between verb attributes, and (3) the semantic distance between subject concepts. The calculation of entity distance is a more in-depth calculation of entities between sentences, including (1) the semantic distance between the entity of the subject and (2) the semantic distance between the entities of the subject, because the semantic distance between the verb attributes is in the concept comparison It has been calculated, so the comparison in the entity is no longer included in the calculation. Finally, the analysis module recommends the workflow based on the conceptual semantic distance and entity semantic distance after nose counting. When making a workflow recommendation, you must make recommendations from small to large according to the size of the semantic distance (the smaller the semantic distance, the smaller the semantic similarity between sentences in Vietnam). The recommended method is ‘(1) Conceptual semantic distance plus

3SS 1227417 V. Description of the invention (4) After the actual semantic distance, the meaning distance, if M 4 ^ recommended '(2) first compare the concept language ^ 丨 丨 the concept of semantic distance is the same, 爯 h 钭 and 浐 ancient ugly Recommended distance from small to large. Re-matching the body buckle In the result feedback stage, the feedback unit will use the working process and the final job of the new talents. The newly-involved query sentence 'is added to the query history record implementation of the electric data map: Push table shows the query sentence semantic comparison query query phrase "Invention Embodiment 1 1 _ Φ. ^ ^ According to the embodiment of the present invention. ^ ^ W electronic source recommends that the system 10 includes a storage device 15 and #: two devices 12, a memory 13, a display device 14, an input device to connect them together. Anyone who is familiar with the technology 11 , CPU 12, memory 13, display 14 and input device 15 can form a mainframe, personal computer, workstation, notebook computer or other computer equipment. The storage device 11 stores the body library ( 〇nt〇1〇gy) ln and query history 112. Memory 13 contains an analysis module 131 containing program instructions and a feedback module 丨 32. The central processing unit 12 is used to load the memory 13 Analysis module 131 and feedback module 132, according to the program deduction order and the data input by the user through the input device 15, perform query semantic comparison and electronic resource recommendation function, Displayed on the display device 14. Ontology is a descriptive conceptual architecture of association between words, in which the 'contains two separate concept tree (Tree c〇ncept), sector

0213-A40050TWF (Nl); A2BA3133; snowball.ptd Page 9 1227417 V. Description of the invention (5) ---------- 2 1. Document 2 2 and meeting 2 3.繁 9 闰 is in the department of the main one. The structure of the dry sound m a is not according to the embodiment of the present invention, Η 1;-. ^ (Instance). Among the six attributes, the second inspection and the two entities hold "214" as the operation attributes, and their inspections " 213 and "are synonymous operation attributes. Two realities: eight = and" Send "" 213 children Business Technology Lab " 21 6. Knife's stop to 2 1 5 and π electric figure 2b are not according to the embodiment of the present invention = only one concept in the document profile 22 H wooden structure does not contain three sub-concepts, of which π is recorded in each profile 222 contains entities "innovation needle a entity. Sub-concepts", meeting ^ ^ Prospective day meeting minutes, 1 2 2 5; early morning winter, ^ meaning ^ know ^ 23 contains entities '' scientific and technical meeting notice, 226 The sub-concept "period", the second report 224 contains the entity "Innovation Prospective Day End-of-Day Execution Report '图 2c Figure shows that according to the original intention, there are two two in the conference concept tree 23:" The tree structure shows two entities (—e) 1 = Nianyi "231, which includes the final review of " coffee " science and technology: Negotiation ^ Figure 3 shows the luxury according to the present invention, ρ Λ. ^ θ ^ ^ Clear example yoke example query sentence history == intent. The query history record 112 has two _, ^, and the workflow, number 312, and its towel contains a total of four strokes 321 to 324 2. The master-pen record stores the past user input query sentences and the workflow of reselection and execution. The query sentence block 311 is used to store the query sentence input in the past, and the work flow number 3 丨 2 block is stored

1227417 V Description of the invention (6) Store the work flow selected and executed by him. Anyone who is familiar with this skill can increase the number of stops in the record or use different fields to name it within the spirit and scope of the present invention. In addition, this record is not limited. Process ID, you can also store the document ID, web page URL, or other electronic resources that can be linked. The electronic resource recommendation of the present invention is divided into two stages. Analysis and result feedback are implemented through the analysis module 131 and the feedback module 132, respectively. The implementation method of these two stages will be described in detail below. The first left analysis module 131 in F is the query sentence entered by the user through the input device 15 and the "planning office sends 92", and the end-of-day execution report ". FIG. 4 is a schematic diagram of an example semantic labeling according to the month 1 of the present invention. 41 to 44 查询 + ^ «ζι + of if 112 t M i.321 ^ 324 Λ V Λ ^ ^ ^ ^ ^ ^ "Analyze, and the 45th line new input analysis module 131" parses out "the main word ^ for historical query sentences in the newly entered sentence history record H2 and the query abbreviation sentence containing time, and The verb (verb) -subject (object) " will be shown in the block sequence of H = bJect). For example, the query sentence "Planning Room Inspection" is freshly analyzed, and the results are as shown in Section 4 7 The execution report "will be analyzed by the enterprise ^^ ... 12-year pre-innovation plan end-of-year execution report" sentence pattern, and the query "" to ~ delivery-the query of innovation prospects 3 2 1 query" planning room The report of "Sixty-one" historical record 11 2 Disciplinary Record "will be interpreted as" planning room-inspection, customs-the annual meeting of the day-to-day innovation forward-looking plan. " 、 4 Prospective day meeting minutes 0213-A40050TWF (Nl); A2BA3133; snowball.ptd Page 11 1227417 ----——________ V. Description of the invention (7) * ------ Medium = down, from the analysis The subsequent "subject-verb-acceptor" sentence sequence sequence is the main word, verb, and acceptor, marked with concepts corresponding to the ontology library 1 1 1 2 3 3 4 4 documents 22 and conference 23 concept tree , Attributes, and two two '^ Note results are shown in column 48. For example, the newly-entered "enter f to ^ submission-innovation forward-looking day-end execution report," is marked as, department general 1 department attribute-the concept of period-end execution report, and query history records,? 3 21 of the "Planning room ~ submission-meeting plan of innovative forward-looking plan, marked department concept-attribute-meeting record concept". Analysis module 131 according to the semantic meaning of the new rotation query,-= for the query In historical record 112, the semantic meanings of all records are counted as two or eight distances. The calculation of semantic distance is divided into two stages, the conceptual distance calculation and the ^ physical distance calculation. The conceptual distance calculation includes (丨) the semantic distance between subject concepts, (2) the semantic distance between verb attributes and (3) the semantic distance of the concept of the word. The entity distance calculation is a more in-depth calculation of the entities between sentences, including (1) the semantic distance between the entity of the subject and (2) ) Semantic distance between subjective bodies, because the semantic distance between the verb attributes has been calculated in the concept comparison, so it is no longer included in the physical comparison. Distance represents the degree of similarity between semantic annotations between sentences. The larger the distance in the concept tree, the lower the similarity (the distance is 0, meaning the same semantic annotations), otherwise the similarity is higher. Semantic distance is defined as a concept or The shortest path on the concept tree between entities, that is, the shortest path related from concept A to concept B, or the shortest path between the entities that belong to the concept. Figure 5 is a schematic diagram of an example semantic similarity calculation according to an embodiment of the present invention. In this example, the query history record 11 2 records 3 2 1 and the newly entered query statement.

0213-A40050TWF (Nl); A2BA3133; snowball.ptd Page 12 I227417

The concepts of the subject and the verb are the same, but the related distance of the subject is 2 (Meeting ^ Record—File—End-of-term executive report), so the conceptual distance is ° + 2 = 2. The query history records 12 and 321 and 3 22 in the historical record 12 and the entity distance of the new query are 4 and 6 respectively. Because the meaning of the words in the record 3 2 3 or 4 is the same as the concept tree in the newly input query sentence, the conceptual distance and the physical distance cannot be compared with

Each last ’analysis module 1 3 1 makes a workflow recommendation based on the calculated conceptual semantic distance and two-body semantic distance. In the process of defamation, we must recommend according to the size of the semantic distance, from small to large (semantic distance 2, the higher the semantic similarity between small representative sentences). The recommended methods are as follows: (/) Conceptual semantic distance plus entity semantic distance, recommended from small to large 2 (2) First, compare the conceptual semantic distance. If you encounter if the conceptual semantic distance is the same as if, open y, then Compare body-to-body semantic distance and recommend it from small to large. In the feedback stage, the feedback unit 132 adds the last job “• process” and the newly entered query sentence by the user to the query history sentence box 丄 1 Z.

FIG. 6 is a flow chart showing a method for semantically comparing query sentences to an electronic resource recommendation method according to an embodiment of the present invention. This method is implemented by code and can be loaded and executed by a central processing unit. First, as in step S611, enter a new query sentence, such as, the planning office :: the 91-year-old innovation forward-looking plan end-of-year execution report. Next, :: step S6 1 2 ′ enter the query history statement record i} 2. Referring to FIG. 3, there are two stops in the history record 1 丨 2 of a π sentence. The query sentence 3 1] and]

393 1227417 V. Description of the invention (9) Process number 312. In step S621, the new query sentence input in step S6 11 and each historical sentence in the historical sentence 112 of the query sentence history record 112 input in step S612 are Chinese-segmented, and parsed out, 'subject-verb-acceptor' The sequence of the sentence pattern is shown in position 47 in Figure 4. Next, in step S622, the new query sentence and the sentence pattern sequence of each historical query sentence are respectively marked as the main word, verb, and subject, and marked with the concepts of department 21, file 22, and meeting 23 in the ontology library 1 11 For the concepts, attributes and entities in the tree, the labeling results are shown in column 4 8 in Figure 4. '

As shown in step S631, the semantic meanings of the newly input query sentence are compared with one another. The semantic meanings of all records in the historical record 112 of the query sentence are compared to calculate the semantic distance. Semantic distance calculation is divided into two stages, conceptual distance calculation and physical distance calculation. The conceptual distance calculation includes (丨) the semantic distance between the subject concepts, (2) the semantic distance between the verb attributes, and (3) the semantic distance between the subject concepts. The calculation of entity distance is a more in-depth calculation of the entities between sentences, including (1) the semantic distance between the entity of the subject and (2) the semantic distance between the entities of the subject, because the semantic distance between the verb attributes is in the concept comparison It has been calculated, so the comparison in the entity is no longer included in the calculation. The calculation results of the conceptual distance and the physical distance are shown as columns 52 and 53 in Fig. 5, respectively. In step S641, a workflow recommendation is performed according to the calculated conceptual semantic distance and entity semantic distance. When making a workflow recommendation, you must make a recommendation from small to large according to the size of the semantic distance (the smaller the semantic distance, the higher the semantic similarity between sentences). There are two recommended methods. (丨)

0213-A40050TW (N1); A2BA3133; snowba 11. p td p.14 394 1227417 V. Description of the invention (10) After the semantic distance is added to the physical semantic distance, it is recommended from small to large; (2) Compare the concepts first Semantic distance. If you encounter a situation where the conceptual semantic distance is the same, then compare the actual semantic distance and recommend it from small to large. Finally, in step S651, the last query executed by the user and the newly entered query statement are added to the query history statement record. Furthermore, the present invention proposes a computer-readable storage medium for storing a computer program. The above computer program is used to implement a spam filtering method with semantic understanding function. This method will perform the steps described above. FIG. 7 is a diagram showing a voice comparison of a query sentence according to an embodiment of the present invention. The following is a schematic diagram of a computer-readable storage medium: 'This storage medium 70 is used to store a computer program 72. The electronic resource recommendation for semantic comparison is recommended. method. The series of electric current query sentences is inputting new query logic 7 2 !, and onggong contains seven logical record logics 722, generating, subject — verb — accept, π = entering query history sentence marking semantic mark logic 724 Calculate the query logic sequence sequence logic 72 3, 725, and generate the semantic distance logic 727 of the electronic resource recommendation list logic. Therefore, by using the search technology resource recommendation system and method provided by the present invention, the accuracy of the electronic comparison of the key ^ 1 δσ sentence-to-sense comparison is more effectively used, and the disadvantage of the word L pair is used more effectively. To increase recommendation. σ 纟 pairing to push electronic resources Although the embodiment of the present invention is only used to limit the present invention, anyone familiar with this technology] ^ 裎 recommends, but it is not a sentence-to-sense comparison method, which is applied to web pages,:, This query file, web service, or

1227417 V. Description of invention (11) Recommendation of other electronic resources. In addition, without departing from the spirit and scope of the present invention, some modifications and retouching can be done, so the protection scope of the present invention shall be determined by the scope of the appended patent application.

0213-A40050TWF (N1); A2BA3133; s nowba11.ptd page 16 1227417 The diagram is briefly explained in order to make the above-mentioned objects, features and advantages of the present invention more comprehensible. The following examples are given in conjunction with the accompanying drawings Detailed description is as follows: FIG. 1 is a system architecture diagram of an electronic resource recommendation system for semantic comparison of query sentences according to an embodiment of the present invention; FIG. 2a is a schematic diagram of a department concept tree architecture according to an embodiment of the present invention. Figure 2b is a schematic diagram of a document concept tree architecture according to an embodiment of the present invention; Figure 2c is a schematic diagram of a conference concept tree architecture according to an embodiment of the present invention; Figure 3 is a schematic query history record of an example query sentence according to an embodiment of the present invention Figure 4 is a schematic diagram of an exemplary semantic tagging according to an embodiment of the present invention, Figure 5 is a schematic diagram of an exemplary semantic similarity calculation according to an embodiment of the present invention; Figure 6 is a diagram illustrating the semantic meaning of a query sentence according to an embodiment of the present invention Method flow chart of the comparison method of the electron negative source recommendation method; the first table shows the semantic meaning of the query sentence according to the embodiment of the present invention. The computer-readable storage medium diagram of the recommended electronic resource comparison method. Explanation of symbols 10 ~ electronic resource recommendation system; 1 storage device; 1 2 ~ central processing unit;

0213-A40050TWF (Nl); A2BA3133; snowball.Ptd 1227417 Schematic description 1 3 ~ Memory; 1 4 ~ Display device; 15 ~ Input device; I 6 ~ Bus bus; 111 ~ Ontology library; II 2 ~ Query statement Historical records; 1 3 1 ~ analysis module; 1 3 2 ~ return module; 2 department concept tree; 2 1 Bu 丨, department 丨 丨 concept; 2 1 2 ~ `` attachment '' attribute; 213 ~ π Check-in "attributes; 214 丨 丨 convention 丨 丨 attributes; 2 1 5 ~ '' Planning room π entity; 2 1 6 ~ π e-commerce technology laboratory '' entity; 2 2 ~ file concept tree; 221 ~ π fileπ Concepts; 2 2 2 ~ 丨 丨 Meetings, 丨 Sub-concepts; 223 ~ 丨, Meeting notifications, 丨 Sub-concepts; 224 ~ 丨, Final execution report 丨 丨 Sub-concepts; 225 ~ 丨 丨 Innovative Prospective Day Meeting Records 丨 丨Entity; 2 2 6 ~ π scientific meeting to notify π entity; 227 ~ π innovation forward-looking plan end-of-term execution report "entity; 2 3 ~ conference concept tree;

0213-A40050TWF (Nl); A2BA3133; snowball.ptd Page 18 1227417 Schematic description of 23 丨 丨 Conference 丨 丨 Concepts; 232 ~ 丨 f Innovation Preview Conference 丨 丨 Entity; 233 ~ 丨 丨 Final Review Conference "Entity; 3 11, 4 6, 5 1 ~ query sentence field; 3 1 2 ~ workflow number field; 41, 42, ..., 45, 51, 52, ..., 54 ~ record 4 7 ~ sentence pattern sequence 48 ~ Semantic mark 5 2 ~ Concept distance 5 3 ~ Entity distance 7 0 ~ Storage media; 7 2 0 ~ Electronic resource recommendation computer program for semantic comparison of query sentence 721 ~ Input new query sentence logic; 722 ~ Enter history query sentence record Logic; 723 ~ generating π subject-verb-acceptor π sentence sequence logic; 7 2 4 ~ tagging semantic tagging logic; 7 2 5 ~ calculating semantic distance logic between query sentences; 726 ~ generating electronic resource recommendation list logic; 727 ~ Store execution result logic.

Page 19 0213-A40050TWF (Nl); A2BA3133; snowball.ptd

OQQ

Claims (1)

1227417 The scope of the six patent applications includes the # Chagou-sentence-comparison electronic resource recommendation system, including a storage device, historical records, the above Taiyuan Temple storage-ontology library, and plural query statements, historical statement storage, and historical record storage. Relevant identifiers of the above-mentioned search sub-resources; and a history query sentence and a corresponding history query query sentence and the first word of the query sentence used by an electric analysis module to turn a history record, from # 麻 士 Γ Sentence, from the above new query statement to generate a set of the corresponding $ first-3434 knife analysis of the above ontology library to calculate ΪΚίίίΐ should be one of the second words above the second semantic mark of one of the above ontology library The correspondence between the mark and the above second semantic mark corresponds to the historical search percentage at the 々 / a tolerance distance. The above analysis module will correspond to the above history. The "Ϊ electronic resource identification" is arranged according to the nearest semantic distance of the above semantic distance. , Displayed on the display device. Electronic 2 Assets: Πί: The query semantic meaning comparison described in the first item of the scope of Deto = ΐ 推 4 which includes-feedback mode ㉟ ' Tap on the above rainbow: Second, store the above-mentioned electronic resource identification code and the new query sentence executed by the user into the history record of the query sentence. 3. Compare the semantic meaning of the query sentence as described in item 1 of the scope of patent application. Electronic resource recommendation system, in which the part of speech of the first word is a main word, verb, or text, and the part of speech of the second word is a main word, verb, or subject. Electronic resource recommendation system for semantic comparison, wherein the first word and the second word are not included 48Q 1227417 Sixth, the scope of patent application --- including the description of time. 5 · The query sentence as described in item 1 of the scope of patent application 谇 咅 electronic resource recommendation system, wherein the electronic resource identification "t pairs of 0213-A40050TWF (Nl); A2BA3133; snowball.ptd Page 21 1227417 Therefore, the semantic meaning comparison system of the query sentence described in item 9 of the scope of the patent application does not include the description of time in the first word and the second word in the above. According to the query semantic comparison system described in item 8 of the scope of the patent application, the above semantic distances in ^ 'eight are conceptual semantic distances, attribute semantic distances, and physical semantic distances. -13. An electronic resource recommendation method for query sentence semantic comparison, which is executed by an electronic device with a central processing unit. The method includes the following steps to input a new query sentence; input a historical query from a query history record; Statement, wherein the history of the query sentence stores at least one historical query sentence and a corresponding electronic resource identifier; a first word is generated from the new query sentence; a second word is generated from the historical query sentence; retrieved by an ontology database Corresponding to the first semantic tag of one of the first words above, wherein the ontology library stores the association structure between words and words; the second ontological root corresponding to one of the second words is retrieved by the ontology library; ~ the above is not calculated The first semantic mark and the second semantic mark correspond to a semantic distance of the ontology library; and the electronic resource identification codes corresponding to the historical query sentence are arranged from near to far according to the approximate degree of the semantic distance, and are displayed in Display device. 》 1 4 · The semantic comparison of the query sentence as described in item 丨 3 of the scope of patent application 1227417 The = ^ sub-resource recommendation method further includes a step of storing the two or two sub-resource identifiers executed by the user and the above query into the previous sentence history. < A slang term 15. The query sentence semantic comparison to Gu Si, Γ resource recommendation method as described in item 13 of the scope of patent application, wherein the first part of speech is part of speech, verb # or text, where the above The part of speech of the second word is the main word, verb, or subject. 16 · The electronic resource recommendation method for the semantic comparison of query sentences as described in item 15 of the scope of patent application, wherein the first word and the second, including the description of time. No 17. The method for recommending electronic resources according to the semantic comparison of query sentences described in item 13 of the scope of patent application, wherein the electronic resource identification code represents a network service, a web page, an electronic workflow or an electronic file. 18. The method for recommending electronic resources for semantic comparison of query sentences as described in item 13 of the scope of patent application, wherein the electronic resource identification code represents a network service, a web page, an electronic workflow or an electronic file. 19 · The electronic resource recommendation method for the semantic comparison of query sentences as described in Item 13 of the scope of patent application, wherein the above semantic distance is the conceptual semantic distance, the attribute semantic distance or the entity semantic distance. 2 0 — A query semantic comparison method executed by an electronic device with a central processing unit. The method includes the following steps: inputting a first query; inputting a second query; from the first query Generate a first word; generate a second word from the second query; 0213-A40050TWF (N1); A2BA3133; snowbal1.ptd page 23 1227417 6. The scope of patent application is retrieved by an ontology library corresponding to the first semantic mark corresponding to one of the above / words, where the above ontology library stores the words between words Relevant architecture; searching the second semantic mark corresponding to one of the second words by the ontology library; and calculating the semantic distance between the first semantic mark and the second semantic mark $ corresponding to one of the ontology libraries. 2 1 · The semantic comparison method of a query sentence as described in Item 20 of the scope of patent application, wherein the part of speech of the first word is the main word, verb, or acceptor, and the part of speech of the second word is the main word, verb, or Subject to words. 2 2 · The method of semantic comparison of query sentences as described in item 21 of the scope of patent application, where the first word and the second word do not include the description of time ° 2 3 · As described in the scope of patent application No. 20 The query semantic comparison method 'where the above semantic distance is conceptual semantic distance, attribute semantic distance or entity semantic distance. 24 · —A kind of 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 the scope of patent applications No. 丨 3 to 丨 9 The method described. 25 · —A kind of computer-readable storage medium for storing a computer program. The computer program is used to load into a computer system and make the computer system execute any one of the patent application scopes 20 to 23. The method described.