KR100898458B1 - Method for searching information and system thereof - Google Patents

Abstract

An information retrieval method and system thereof are provided. The information retrieval method includes analyzing types of a plurality of queries, classifying queries according to the analyzed types, indexing information using the classified queries, and using the index results from a user. Retrieving information corresponding to the entered query, wherein the type for the query includes a category, brand, model name, or synonym query.

Category, feedback information, search

Description

Method for searching information and system

The present invention relates to an information retrieval method and system, and more particularly, to an information retrieval method and system for retrieving information suitable for the user's query intent.

In general, an information retrieval method analyzes a query input by a user and provides search results corresponding to the analyzed query in order of popularity.

For example, if a user enters 'drum' as a query and wants to search for information related to the drum, which is a musical instrument, the conventional information retrieval method is a search result corresponding to 'drum' by another user. If high, the drum washing machine preferentially provides a search result over a drum, which is an instrument to be searched, so that the user cannot preferentially provide a desired search result.

For example, if the query entered by the user is 'jeans', the conventional information retrieval method searches only the popular items among men's jeans or women's jeans, so if the user wants women's or men's jeans, only the popular products among women's jeans are searched. Overprovisioning may result in a user not properly searching for the desired information.

For example, when the query input by the user is 'puma', the conventional information retrieval method is actually used when the search results including 'Nike / Adidas / Polo / Beanpole' are exposed to the top by Abuse. There is a problem in that it does not provide an accurate search result for the desired puma.

Therefore, there is an urgent need for a method that can grasp the query intention of the user and provide a more accurate information search result desired by the user.

The present invention provides an information retrieval method and system for retrieving information suitable for a user's query intent.

In addition, the present invention provides an information retrieval method and system that removes the abuse to provide a more accurate information search results desired by the user.

In addition, the present invention provides an information retrieval method and system that can improve the manual search results when retrieving information.

The present invention also provides an information retrieval method and system capable of reflecting a merchandise sales degree in a search ranking.

The present invention also provides an information retrieval method and system that can reflect the product release date in the search ranking.

An information retrieval method according to an embodiment of the present invention includes analyzing types of a plurality of queries, classifying a query according to the analyzed types, and indexing information using the classified queries. And retrieving information corresponding to a query input from a user using the index result, wherein the type of the query includes a category, a brand, a model name, or a synonym query.

An information retrieval system according to an embodiment of the present invention, the analysis unit for analyzing the types of the plurality of queries, the classification unit for classifying the query according to the analyzed type, and the information using the classified query An index unit for indexing and a search unit for searching for information corresponding to a query input from a user using the index result, the type of the query includes a category, brand, model name or synonym query.

According to the present invention, an information retrieval method and system for retrieving information suitable for a user's query intent can be provided.

In addition, according to the present invention, it is possible to provide an information retrieval method and system that provides a more accurate information retrieval result desired by the user by eliminating the abuse.

In addition, according to the present invention, it is possible to provide an information retrieval method and system that can improve the manual search results when retrieving information.

In addition, according to the present invention, it is possible to provide an information retrieval method and system that can reflect the product sales degree in the search ranking.

In addition, according to the present invention, it is possible to provide an information retrieval method and system that can reflect the product release date in the search ranking.

Hereinafter, an information retrieval method and system thereof according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating an interlocking relationship between an information retrieval system and a user terminal according to the present invention.

Referring to FIG. 1, when a user terminal 110 receives a query to search from a user, the user terminal 110 transmits the input query to the information retrieval system 120.

The information retrieval system 120 analyzes the types of the plurality of queries stored in the database. The type for the query may include a category, brand, model name or synonym query. That is, the information retrieval system 120 analyzes the query intention using the feedback information and ranks the category in which the information in which the query is most clicked according to the analysis result of the query intention is selected. Can be analyzed. The feedback information may include a click frequency for a query.

The information retrieval system 120 may remove the duplicate category name when the category for the information overlaps the upper category name and the lower category name according to the analyzed type. In addition, the information retrieval system 120 may remove a category including the keyword except for the predetermined category when the predetermined keyword is included in the category for the information according to the analyzed type. In addition, the information retrieval system 120 may remove the feedback information when the feedback information has a click frequency less than or equal to a threshold value according to the analyzed type.

The information retrieval system 120 may rank the information using the section having the highest section weight by applying section weights to categories, brands, manufacturers, model names, keywords, or feedback information according to the analysis result of the query intention. Can be. In addition, the information retrieval system 120 may analyze the type of the query by reflecting the cumulative click frequency for the query, the Abyss reading index, or the sales index for each product. The information retrieval system 120 may analyze the type of the query by applying penalties to the accumulated click frequency, the Abyss reading index, and the sales index, when the Abyss reading index is less than or equal to a reference value.

The information retrieval system 120 classifies the query according to the analyzed type, and indexes the information using the classified query. The information retrieval system 120 may expand the query by adding synonyms to the query using the synonym dictionary.

The information retrieval system 120 retrieves information corresponding to a query input from a user using the index result. In addition, the information retrieval system 120 provides the retrieved information to the user terminal 110.

2 is a diagram illustrating a configuration of an information retrieval system according to an embodiment of the present invention.

2, the information retrieval system 120 includes an information database 210, an analysis unit 220, a classification unit 230, an index unit 240, an index database 250, and a search unit 260. do.

The information database 210 records and maintains a plurality of queries input by users and various information that can be provided according to the queries.

The analyzer 220 analyzes the types of the plurality of queries. That is, the analyzer 220 may analyze the types of the plurality of queries recorded in the information database 210. The type for the query may include a category, brand, model name or synonym query.

The analysis unit 220 analyzes the query intention using the feedback information, and analyzes the information by ranking the category in which the information in which the query is clicked the most is located according to the analysis result of the query intention. The feedback information is a user query set that clicks on information, and means for identifying the search intention of users, and a click indicating a query used when the information is clicked and the number of times the information is clicked by users using the query. Include frequency.

5 is a diagram illustrating an example of feedback information.

Referring to FIG. 5, the product information 510 represents adidas jersey sweatshirt warm set sweats (black / white) as shopping product information clicked by the user among search results for a query input from a user. The feedback information 520 is a set of user queries that click on the product information 510. The feedback information 520 indicates a click frequency indicating the number of times the product information 510 is clicked by the users and when the product information 510 is clicked. Contains the entered query. For example, the feedback information 520 may have a click frequency of 412 when the query is a sweat suit, a click frequency of 259 when the query is adidas, a click frequency of 196 when the query is adidas sweatshirt, and the query. Is 182 when the sportswear, and the click frequency is 160 when the query is adidas sportswear.

As such, the analysis unit 220 analyzes the query intention using the feedback information as shown in FIG. 5, and ranks the category in which the information in which the query is clicked most exists according to the analysis result of the query intention. can do.

Also, the analysis unit 220 ranks the information using the section having the highest section weight by applying section weights to categories, brands, manufacturers, model names, keywords, or feedback information according to the analysis result of the query intention. can do.

3 is a diagram illustrating an example of a specific configuration of an analysis unit.

Referring to FIG. 3, the analyzer 220 includes a category refiner 310, a feedback information refiner 320, and a popularity applier 330.

When the category for the information overlaps the upper category name and the lower category name according to the analyzed type, the category refining unit 310 purifies the category by removing the duplicate category name. That is, in the case of a multi-category category name, the category refining unit 310 may refine the category by removing the duplicate upper category name if the upper category name overlaps with the lower category name.

6 is a diagram illustrating an example of refining a category.

Referring to FIG. 6, when the first category 610 is refined, the category refiner 310 may have a 'notebook' as an upper category name and a lower category name such as 'notebook / desktop / monitor> notebook> LG electronic'. As it is duplicated, duplicate parent category name 'laptop / desktop / monitor' can be removed.

For example, in the case of refining the second category 620, the category refining unit 310 is synonymous with the upper category name 'dica' and the lower category name 'digital camera' such as 'dica / camcorder> digital camera> canon'. Therefore, since it is substantially duplicated, it is possible to remove the duplicate category name 'dica / camcorder'.

For example, when refining the third category 630, the category refining unit 310 does not remove the category name because the upper category name does not exist in the lower category name, such as 'diaper / sanitary products> panty type> Huggies', Diapers / Hygiene items> Panties> Huggies' can be used as it is.

Also, when the category refining unit 310 includes a keyword having an implicit meaning in a category, the category refining unit 310 selects a specific meaning and removes the keyword having the implicit meaning from all categories except the category including the keyword having the specific meaning. can do. For example, when the category includes the keyword 'guitar', the category is maintained for the category 'guitar' which is an instrument, and the category 'guitar' in all category names except the category for the guitar 'guitar'. Can be removed.

The feedback information refiner 320 purifies the feedback information by removing the feedback information when the feedback information has a click frequency less than or equal to a threshold value according to the analyzed type. That is, the feedback information refiner 320 may refine the feedback information by defining a click frequency threshold in order to refine the accumulated feedback information for each information, and removing feedback information having a click frequency less than or equal to the threshold. The click frequency threshold may be determined with reference to a click frequency value, preferably a maximum click frequency value, in feedback information.

7 is a diagram illustrating an example in which feedback information is purified.

Referring to FIG. 7, the click frequency threshold may be set to '40' as an example by referring to the maximum click frequency in the feedback information 700. The feedback information refiner 320 may remove the Lee Hyo Lee phone having a click frequency of '25' and the Sch-S470 having '10' from the feedback information 700.

The popularity application unit 330 applies the accumulated click frequency, the Abuse reading index, or the sales index for each product according to the analysis result of the query intention as the popularity. That is, the popularity application unit 330 may log-normalize the accumulated click frequency, normalize the Abuse judgment index of the brand name through morphological analysis, and apply the popularity using the maximum value in the list of the product's sales index. Can be. In this case, the popularity application unit 330 may reflect the freshness score of the information generation period (Maxage) within 6 months to the popularity. In addition, the popularity application unit 330 may apply a penalty to the popularity of the cumulative click frequency, the Avues reading index and the sales index, when the Abeus reading index is less than or equal to the reference value. As an example, when the ABBIS reading index is less than or equal to a reference value, the popularity applicator 330 may apply a penalty to the popularity by reducing the click frequency of the accumulated click frequency, the ABBUS reading index, the sales index, and the feedback information. have.

The analyzer 220 may define a search ranking by combining the analyzed query intent, popularity, and similarity. For example, the analyzer 220 may define a similarity using the product name (model name), category name, brand name, and frequency of feedback information.

The classification unit 230 classifies the query according to the analyzed type.

4 is a diagram illustrating an example of a configuration of a classification unit.

Referring to FIG. 4, the classifier 230 includes a query extension unit 410 and a query classifier 420. The query extension unit 410 expands the query by adding a synonym for the query using a synonym dictionary. That is, the query extension unit 410 expands the query by applying an ORMAX operation using the synonym dictionary. The extended query is a concept, and the number of queries is one. In addition, the query extension unit 410 does not generate similarity by adding weights when merging posts, but uses a maximum value and does not divide the number of queries even in the process of normalizing the weights. For example, when the query is a 'mobile phone', the query extension unit 410 may extend the query to a 'mobile phone, a mobile phone, a new phone, a handfun' and the like by using the synonym dictionary.

The query classifier 420 classifies the extended query.

The indexing unit 240 indexes the information using the classified query.

Index database 250 records and maintains the index results. The index result is the query and information corresponding to the query.

The search unit 260 searches for information corresponding to a query input from a user using the index result. That is, the search unit 260 may search the information corresponding to the query input from the user with reference to the index database 250.

As such, the information retrieval system according to the present invention can retrieve information of a category suitable for a user's query intent.

8 is a diagram illustrating a flow of an information retrieval method according to an embodiment of the present invention.

Referring to FIG. 8, in operation S810, the information retrieval system analyzes types of a plurality of queries. The type for the query may include a category, brand, model name or synonym query.

9 is a diagram illustrating an example of a more detailed step of analyzing a type of a query.

Referring to FIG. 9, in step S910, the information retrieval system analyzes query intention using feedback information. The feedback information may include a click frequency indicating a number of clicks of the query and product information clicked by users who use the query.

In step S920, the information retrieval system ranks the category in which the information in which the query is clicked most exists according to the analysis result of the query intention. That is, in step S920, the information retrieval system uses the section having the highest section weight by applying section weight to category, brand, manufacturer, model name, keyword, or feedback information according to the analysis result of the query intention. To rank the information.

For example, in operation S810, when the category for the query overlaps the upper category name and the lower category name according to a result of analyzing the type of the query, the information retrieval system may remove the duplicate category name.

For example, in operation S810, when the feedback information has a click frequency less than or equal to a threshold value according to a result of analyzing the type of the query, the information retrieval system may remove the feedback information.

For example, in step S810, the information retrieval system may analyze the type of the query by reflecting the cumulative click frequency for the query, the Abyss reading index, or the sales index for each product.

For example, in operation S810, when the Abbers reading index is less than or equal to a reference value, the information retrieval system may analyze a type of the query by applying a penalty to the accumulated click frequency, the Abbuzz reading index, and the sales index. have.

In step S820, the information retrieval system classifies the query according to the analyzed type. For example, in operation S820, the information retrieval system may expand the query by adding a synonym for the query using a synonym dictionary, and then classify the extended query according to the analyzed type.

In step S830, the information retrieval system indexes the information using the classified query.

In operation S840, the information retrieval system retrieves information corresponding to a query input from a user using the index result.

As described above, the information retrieval method according to the present invention can retrieve information of a category suitable for the query intention of the user.

Meanwhile, the information retrieval method according to the present invention can be implemented in the form of program instructions that can be executed by various computer means and recorded in a computer readable recording medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

1 is a diagram illustrating an interlocking relationship between an information retrieval system and a user terminal according to the present invention.

2 is a diagram illustrating a configuration of an information retrieval system according to an embodiment of the present invention.

3 is a diagram illustrating an example of a specific configuration of an analysis unit.

4 is a diagram illustrating an example of a configuration of a classification unit.

5 is a diagram illustrating an example of feedback information.

6 is a diagram illustrating an example of refining a category.

7 is a diagram illustrating an example in which feedback information is purified.

8 is a diagram illustrating a flow of an information retrieval method according to an embodiment of the present invention.

9 is a diagram illustrating an example of a more detailed step of analyzing a type of a query.

<Explanation of symbols for the main parts of the drawings>

110: user terminal 120: information retrieval system

210: information database 220: analysis unit

230: classification unit 240: index unit

250: index database 260: search unit

310: category refiner 320: feedback information refiner

330: popularity application

410: query extension unit 420: query classification unit

Claims (19)

delete Analyzing the types of the plurality of queries; Classifying the query according to the analyzed type; Indexing information using the classified query; And Searching for information corresponding to a query input from a user by using the index result Including, The type for the query is Include category, brand, model name, or synonym query, The step of analyzing the type for the query, Analyzing the query intention using the feedback information; And Ranking a category in which the information in which the query is clicked most is located according to an analysis result of the query intention Including, The feedback information, An information retrieval method including a click frequency for the query. The method of claim 2, The step of analyzing the type for the query, And removing the duplicate category name when the category for the query is duplicated with the upper category name and the lower category name according to the result of analyzing the type of the query. The method of claim 3, The step of removing the duplicate category name, And when the predetermined keyword is included in the category for the query, removing the keyword from all categories except the predetermined category. The method of claim 2, The step of analyzing the type for the query, And removing the feedback information when the feedback information has a click frequency less than or equal to a threshold value according to a result of analyzing the type of the query. The method of claim 2, The step of ranking the category, And ranking the information using the section having the highest section weight by applying section weights to categories, brands, manufacturers, model names, keywords, or feedback information according to analysis results of the query intent. The method of claim 2, The step of analyzing the type for the query, An information retrieval method for analyzing the type of the query by reflecting the accumulated frequency of clicks on the query, the Abyss reading index, or the sales index for each product. The method of claim 7, wherein The step of analyzing the type for the query, And analyzing the type of the query by applying a penalty to the accumulated click frequency, the Abuse reading index, and the sales index, when the Absolute Reading Index is less than or equal to a reference value. The method of claim 2, And expanding the query by adding synonyms to the query using a synonym dictionary. A computer-readable recording medium in which a program for executing the method of any one of claims 2 to 9 is recorded. delete An analysis unit for analyzing types of a plurality of queries; A classification unit classifying the query according to the analyzed type; An indexing unit for indexing information using the classified query; And Search unit for searching the information corresponding to the query input from the user by using the index results Including, The type for the query is Include category, brand, model name, or synonym query, The analysis unit, Analyze the query intention using feedback information, and analyze the types of the plurality of queries by ranking the category in which the information in which the query is clicked the most is located according to the analysis result of the query intention, The feedback information, And a click frequency indicative of the query and the number of times the query was clicked by users. The method of claim 12, The analysis unit, And a category refining unit for refining the category by removing the duplicate category name when the category for the information overlaps the upper category name and the lower category name according to the analyzed type. The method of claim 13, The category refiner, And a predetermined keyword is included in a category for the information according to the analyzed type. The method of claim 12, The analysis unit, And a feedback information refining unit for refining the feedback information by removing the feedback information when the feedback information has a click frequency below a threshold value according to the analyzed type. The method of claim 12, The analysis unit, And ranking the information using the section having the highest section weight by applying section weights to categories, brands, manufacturers, model names, keywords, or feedback information according to analysis results of the query intent. The method of claim 12, The analysis unit, And a popularity application unit configured to apply the accumulated click frequency, the Abyss reading index, or the sales index for each product as popularity according to the analysis result of the query intention. The method of claim 17, The popularity application unit, And the penalty is applied to the popularity for the cumulative click frequency, the avues reading index, and the sales index, when the avails reading index is less than or equal to a reference value. The method of claim 12, And a query extension unit for expanding the query by adding synonyms to the query using a synonym dictionary.

Images