KR20090036918A - Method and system for serving document exploration service based on title clustering - Google Patents

Abstract

A method and a system for serving document exploration service based on title clustering are provided to efficiently search a document by mapping a title cluster formed based on a title of a document to directory corresponding to the subject. A document classification unit(211) classifies a document according to theme. A title extracting unit(212) extracts a title of a document. A document comprises one or a plurality of field. The title extracting unit extracts a title in consideration of an attribute of a field comprising document. A cluster forming unit(213) forms a cluster based on the extracted title. A directory mapping unit(214) maps a cluster in a directory belonging to theme. A document delivery unit(215) provides a document to a user terminal(230). The document delivery unit provides a visualization unit. The structure of cluster and the directory in which document belongs is visualized to visualization unit.

Description

Method and system for serving document exploration service based on title clustering}

The present invention relates to a method and system for providing a document search service based on title clustering.

There are numerous documents on the web about various interests. Users can obtain information by delivering query information about the information they want to a search engine. However, entering a query on a topic of interest each time is very cumbersome.

Meanwhile, in order to access information desired by the user without accessing a search word, a user may access a vertical site and a blog having expertise in a specific field to obtain the latest information of the corresponding field.

The level of information existing in these vertical sites and blogs is improving day by day, and is developing as the media that can acquire the fastest and deepest information in the field.

However, visiting each site to view information scattered across multiple vertical sites and blogs can also be inconvenient for the user. As a supplement, vertical sites and blogs provide Really Simple Syndication Feeds (RSS Feeds), and programs such as RSS Readers can be used to subscribe to them.

However, each RSS feed provides information independently of each other, and even if the document has the same or extremely similar contents, it is treated as separate information. Therefore, an additional effort to improve efficiency in the process of searching and viewing the information is needed. Required.

The present invention provides a method and system for providing a more efficient document search service by classifying documents by subject and mapping a cluster formed based on the title of the document to a directory belonging to the subject.

In addition, the present invention provides a method and system for providing a document search service that classifies documents collected on the web and visualizes the clusters formed based on the titles of the classified documents and the directories to which the clusters are mapped, thereby reconstructing the efficiency of information search. It is.

According to an aspect of the present invention, the method comprises: classifying documents according to a subject; Extracting a title of the document; There is provided a method for providing a document search service including forming a cluster based on the extracted title and mapping the cluster to a predetermined directory belonging to the subject.

The method for providing a document search service may further include providing a document to a user terminal, and the document providing step may provide a means for visualizing a structure of a directory and a cluster to which the document belongs.

In the document search service providing method, the document may include one or a plurality of fields, and the extracting the title of the document may extract the title in consideration of the attributes of the fields constituting the document.

Meanwhile, in the forming of the cluster, the extracted title may be divided into syllable units, and a part of the title shared with other documents may be selected as a candidate candidate for cluster concept. Core concept candidates may be selected using n-gram analysis of the title of the document.

In the forming of the cluster, the cluster may be formed when there are more than a predetermined number of documents sharing the central concept candidate. In the case where there are a plurality of central concept candidates, the central concept may be selected in consideration of the length of the title candidate phrase.

Further, according to another aspect of the invention, the document classification unit for classifying documents according to the subject; A title extractor for extracting a title of the document; There is provided a document search service providing system including a cluster forming unit for forming a cluster based on the extracted title, and a directory mapping unit for mapping the cluster to a predetermined directory belonging to the subject.

In addition, the document search service providing system may further include a document providing unit for providing a document to the user terminal, and the document providing unit may provide a means for visualizing the structure of the directory and cluster to which the document belongs.

In a document search service providing system, a document may include one or a plurality of fields, and the title extractor may extract a title in consideration of attributes of fields constituting the document.

Meanwhile, the cluster forming unit may separate the extracted title into syllable units, and select a part of the title shared with other documents as the central concept candidate of the cluster. Core concept candidates may be selected using n-gram analysis of the title of the document.

The cluster forming unit may be configured to form a cluster when there are more than a predetermined number of documents sharing the central concept candidate, and when there are a plurality of central concept candidates, the central concept may be selected in consideration of the length of the title candidate phrase.

Meanwhile, the method for providing a document search service may be performed by a computer, and may be recorded in a computer readable recording medium for recording a program to be executed in the computer.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims and detailed description of the invention.

According to a preferred embodiment of the present invention, a method and system for providing a more efficient document search service can be implemented by mapping a cluster formed based on a title of a document to a directory belonging to a corresponding subject.

Further, according to a preferred embodiment of the present invention, a document search service that classifies documents collected on the web and visualizes the efficiency of information search by visualizing clusters and clusters to which the clusters are mapped based on the classified document titles. Provisioning methods and systems can be implemented.

Hereinafter, an embodiment of a method and system for providing a document search service based on title clustering according to the present invention will be described in detail with reference to the accompanying drawings. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all transformations, equivalents, and substitutes included in the spirit and scope of the present invention. In the following description of the present invention, if it is determined that the detailed description of the related known technology may obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, in the description with reference to the accompanying drawings, the same or corresponding components will be given the same reference numerals and redundant description thereof will be omitted.

1 is a flowchart of a method for providing a document search service according to an embodiment of the present invention, and FIG. 2 is a block diagram of a system for providing a document search service according to an embodiment of the present invention.

1 and 2, the search service providing server 210, the document classifying unit 211, the title extracting unit 212, the cluster forming unit 213, the directory mapping unit 214, and the document providing unit 215. ), An advertisement providing unit 216, an original document database 221, a search service database 222, an advertisement database 223, and a user terminal 230 are illustrated.

Classifying documents according to a subject (S110) is classifying documents in the original document database 221 by the document classifying unit 211.

The document classification unit 211 obtains information about the document from the original document database 221, and obtains information about the classification from the search service database 222. Based on the obtained information, it is determined which classification the document matches, and information about the matching relationship between the document and the classification is stored in the search service database 222.

In this step, the document classifying unit 211 may classify the document by subject by determining whether the document includes a specific keyword and whether the document includes specific content, using information included in the document.

For example, whether the document is a suitable document that matches the classification "wine" includes whether the document contains the classification name "wine" itself and a synonym for the classification name, and keywords that may be considered to have a close relationship with "wine". It may be determined in consideration of whether it includes 'sommelier', 'decanting', and the like.

On the other hand, in determining whether to match the document and the classification, whether or not to include the keyword related to the classification can be used as a reference. For example, if keywords related to a specific subject are included, a predetermined score may be given, and if the sum of these scores exceeds a predetermined criterion, the score may be determined to match the corresponding classification.

In the document classification stage, one document is not necessarily determined to match only one classification. For example, if there is a classification of wine and a Japanese manga, the document dealing with the appreciation of the Japanese comic book, “Drop of God,” which deals with wine, is not only a category for wine, but also “Japan”. May be simultaneously matched to the category 'manga'.

The original document database 221 stores information about original documents that can be classified and reconstructed by the search service providing server 210. These original documents may ultimately be provided to the user terminal 230. The original document may be collected by a web robot or the like on the web.

Meanwhile, original documents stored in the original document database 221 of the present invention may include documents having predetermined attributes. For example, structured documents used in vertical sites and blogs may be used as original documents.

Such structured documents can divide and store content within a document into one or a plurality of regions or sections. Such regions or compartments may be named fields. For example, a document such as a post of a blog may include a title field, a body field, a creation time field, and a keyword field for the post.

With respect to such a document, the creator can generate a document by inputting contents corresponding to each field name, so that the field name and the corresponding content can be usefully used in a title extraction step and the like described later.

Also, in such vertical sites and blogs, the relationship between documents may also be structured. The structured relationships of these documents can appear in the form of directories on the site.

For example, a vertical site about a movie is a directory that classifies documents on the site, and may include directories such as 'movie review', 'movie ranking', and 'latest release', and blogs also categorize each post. It can have information about the directory it plays.

Directory names in these vertical sites and blogs can be used as keywords associated with the topics the site deals with. These keywords can be used in the thematic classification stages of the aforementioned documents and used to improve the accuracy of the classification.

In the present application, the term document may be understood as a term for electronically recorded documents. A document may be described using a markup language such as HTML, and may have an extension of * .htm, but is not interpreted as being limited to a file having a specific description form and extension.

The search service database 222 stores information about a matching relationship between a document and a classification determined by the document classification unit 211. Matching for each category may be stored for each document, and whether or not to include related keywords for each category may be numerically stored.

On the other hand, the information storage method and form in the original document database 221 and the search service database 222 described above may be variously changed within the scope of the present invention.

Extracting the titles of the documents (S120) is a step in which the title extractor 212 extracts the titles of the documents stored in the original document database 221. The title of a document means a word, phrase, or sentence that includes the content and subject of the document.

The title extractor 212 extracts a title of each document by using the information of the document stored in the original document database 221 and stores the extracted title in the search service database 222.

In this step, the title extractor 212 may extract the title of the document using the information included in the documents. The structure of the document, the frequency of occurrence of words included in the document, and attributes when the document is browsed in the user terminal 230 may be used as a criterion for determining the title.

That is, the document information used in the process of extracting the title may be understood as a concept including not only the content text directly included in the document, but also information about the form in which the document is viewed in the user terminal 230.

For example, a web site such as a blog may include structured documents. These documents can store information in fields with their respective names. Text included in a field having a field name such as 'title' or '(title)' may be selected as a title.

As another example, when a document is browsed through a web browser or the like of the user terminal 230, text that is represented and emphasized with a relatively larger or differentiating attribute than other contents in the document may also be considered as a candidate phrase of a title. .

The step S130 of forming a cluster of documents based on the extracted title is a step in which the cluster forming unit 213 performs clustering based on the title information of the document.

The cluster forming unit 213 forms a cluster of documents based on the title information of the documents obtained from the search service database 222. Information about the formed cluster is stored in the search service database 222.

A cluster of documents refers to a group of documents that share a central concept. The cluster of documents may be formed in consideration of whether there are common parts in the titles of each document. Each cluster can be named using a central concept.

A string in common with other documents in the title of the document may be a candidate for the central concept of the cluster. When the number of documents having the common string is equal to or larger than a predetermined value, one independent cluster may be formed.

For example, if the title of one document is 'Catching Sommelier: Wine Etiquette-Wines Enjoying Together' and the other document is titled 'Table Manor 5-Wine Etiquette', the common part of the two titles is 'Wine Etiquette'. Can be extracted as a concept.

An n-gram analysis method may be used in the process of extracting overlapping parts of a document's title as a central concept. In this case, the title may be separated into syllable units and recombined into a string having a predetermined number of syllables.

Duplicate parts of the recombined string may be candidates for the central concept. In the case of the previous example, 'wine' with two syllables and 'wine etiquette' with five syllables can be considered as candidates for the central concept.

As such, when there are several identical overlapping parts in the titles of documents, a process of determining one overlapping part as a central concept may be required. At this time, the number of syllables of the duplicated portion, the relationship between the duplicated portion and the classification name of the document, and the number of documents having the corresponding duplicated portion may be used as the determination criteria.

In the above example, the overlapping portions are 'wine', 'etiquette' and 'wine etiquette'. In this case, 'wine' is the same as 'wine', which is a subject that includes documents, so it may not be appropriate as a central concept of a cluster.

In terms of the number of documents having duplicate portions, the number of documents sharing 'wine' may be too large to form a cluster. As such, when determining the candidate for the cluster-centric concept, it may be required to limit the number of documents sharing the candidate for the central concept to a predetermined range.

In addition, the length of the central concept candidate may also be considered. In the case of an overly short central concept candidate, the part-of-speech may be a general term that is inappropriate for investigation or for a particular classification.

On the other hand, in case of a long concept candidate, it is expected that the relation among documents in the cluster sharing the candidate is high and the probability that noise may be included is low. Can be considered

In the above example, since the remaining two central concept candidates are included in 'wine etiquette', the longest 'wine etiquette' may be considered as a candidate and the other criteria such as the number of documents sharing 'wine etiquette' may be satisfied. If it is determined that it can be selected as a central concept constituting a single cluster.

Also, if the number of documents sharing 'etiquette' and the number of documents sharing 'wine etiquette' are extremely similar in the documents included in the subject of 'wine', it is effective to select more specific 'wine etiquette' as the central concept. Can be.

Among these central concepts, a cluster may be configured based on a predetermined number of documents associated with each other. Information about the central concept constituting the cluster and documents belonging to the cluster that share the central concept is stored in the search service database 222.

The step S140 of mapping a cluster of documents to a directory is a step in which the directory mapping unit 214 maps a cluster of documents to a directory based on a central concept of each cluster.

The directory refers to the classification of documents, that is, subtopics that can include one or more clusters as sub-concepts of the subject. For example, if the subject of a document is categorized as wine, the directory may include wine production, wine history, and wine etiquette. The wine production directory is known as the wine producing region. Bordeaux 'and' Burgundy 'and the like may include a cluster formed around the central concept.

The directory mapping unit 214 obtains information about a directory structure and information about a cluster from the search service database 222 and determines a cluster to be mapped to each directory. The determined directory mapping information is stored in the search service database 222.

The directory to which the cluster is mapped may be determined by whether the central concept of the cluster includes keywords related to the directory.

For example, when the directory is 'wine etiquette', 'etiquette' may be a keyword for determining whether to include a directory in the directory name. In the process of mapping clusters formed for documents that have already been determined to belong to the 'wine' classification, it may be efficient to map the directory with keywords excluding the wine name itself.

On the other hand, for these keywords it may be required to expand the keyword using lexicographical ordering. 'Etiquette' can be expanded with keywords that differ in synonyms, synonyms, and notation.

In this case, keywords such as 'etiquette', 'etiquette', 'manner' and 'manner' may be considered as additional keywords for directory mapping. This can improve directory mapping efficiency.

Keywords for this directory mapping may also be stored in the search service database 222 as part of the information about the directory structure.

In the providing of the document (S150), the document providing unit 215 provides the clustered documents to the user terminal 230 for each directory.

This step is characterized in that the user can easily search for documents of interest by visualizing and providing a predetermined theme, that is, the directory structure belonging to the classification and the inclusion relation of the cluster belonging to the directory.

The user transmits information about a topic of interest to the search service providing server 210 through the user terminal 230. This may be performed by clicking a link to a web page that provides a search service for the corresponding subject.

The document provider 215 receives a request from the user terminal 230, which contains information about a subject of interest of the user, and in response thereto access links to the documents classified and clustered in the aforementioned steps. It can provide a web page including a. This provides a navigation service to the user.

Meanwhile, in order to provide a document search service, the document provider 215 obtains information about clustered documents from the search service database 222.

The form and form in which the response of the document providing unit 215 is provided may be adjusted by using cascading style sheets (CSS). It is also possible to use a separate content management system (CMS).

Meanwhile, the search service providing server 210 according to an embodiment of the present invention may further include an advertisement providing unit 216.

The advertisement provider 216 may provide the advertisement content to the user terminal 230. The advertisement content may be stored in the advertisement database 223 and may be called by the advertisement providing unit 216 and transmitted to the user terminal 230.

As an element for determining advertisement content to be transmitted to the user terminal 230, information about a user and information about a document searched by the user may be considered.

For example, when a user performs a log-in process while using a document search service, personal information of the user, such as the user's age, occupation, gender, and region of residence, may be considered as an advertisement content determining factor. .

Meanwhile, information about a document read through the user terminal 230 may be considered as an advertisement content determining element. Query information entered by the user may also be considered.

In addition, the information obtained while the user uses the document search service according to an embodiment of the present invention may also be considered as a content determining element.

As such, by determining the advertisement content to be provided using information about the user and information about the document searched by the user, there is an advantage that the effect of the provided advertisement can be maximized.

The document search service providing screen according to an embodiment of the present invention will be described in more detail with reference to FIGS. 4 and 5.

On the other hand, the components shown in Figure 2 do not necessarily have a hardware configuration, some components may be implemented in the form of an application program that performs the same or similar functions. In addition, each component may be combined or separated within the scope of the invention.

In the present embodiment, the search service providing server 210, the original document database 221, and the search service database 222 are each configured in an independent form, but may be grouped into one or more elements within the scope of the present invention. have.

In this embodiment, the search service providing server 210, the original document database 221, the search service database 222, and the user terminal 230 may exchange information using a network.

These networks do not necessarily have to be one single network. In addition, the network may be configured in the form of LAN and WAN by technologies such as ADSL, VDSL, Wi-Fi, WIBRO, and HSDPA, and technologies such as VPN may be used to enhance security.

3 is a diagram illustrating a cluster and directory mapping of a document according to an embodiment of the present invention.

In the present embodiment, the documents belonging to the original document database 221 may be structured by a hierarchical structure connected in the order of topics, directories, clusters, and documents. By structuring the documents, information scattered among various vertical sites can be easily searched with a single page view provided by the document search service according to the present embodiment.

Referring to FIG. 3, the classification, which is a higher concept in which documents are structured, is 'wine'. It is determined by the document classifying unit 211 whether a specific document corresponds to a 'wine' classification.

A classification may include one or more directories as a sub-concept. The 'wine' classification includes directories named 'Wine Mountain,' 'Wine History,' and 'Wine Etiquette.'

Since the name of the directory functions as the name of the group that the user wants to search for the document, the user's document is used by using the name of the document group used in vertical sites and blogs, which are the sources of documents stored in the original document database 221, as the directory name. The search efficiency can be improved.

The directory may include one or more clusters below it. The 'mountain of wine' directory includes clusters named 'Bordeaux' and 'Burgundy'. The cluster may also include one or a plurality of documents, and the document 3 illustrated in FIG. 3 is included in both the 'Bordeaux' cluster and the 'Burgundy' cluster.

On the other hand, the names of directories used in vertical sites and blogs and field information of structured documents included in such sites can be used to categorize and cluster documents by subject, as discussed in the detailed description of FIGS. 1 and 2. As shown.

4 is a diagram illustrating a document search service providing screen according to an embodiment of the present invention. Referring to FIG. 4, a classification display area 400, a classification structure display area 410, a search position display area 420, and a cluster display area 430 are illustrated.

As described above, in order to provide the document search service according to the present embodiment, the document provider 215 may generate and transmit a web page provided to the user terminal 230.

The web page provided in the document search service may have a screen configuration as illustrated in FIG. 4. Such a screen configuration may include a display area for visualizing the structure of the searched document.

The classification display area 400 is an area where information on classification, which is a higher concept in which a document is structured, is displayed. In this embodiment, the classification name 'wine' is displayed as a relatively differentiated attribute.

The classification structure display area 410 is an area for providing a user with a classification structure similar to that illustrated in FIG. 3. In the case of FIG. 4, 'the mountain of wine' and the cluster 'Bordeau', which are the directories that the user is searching, are displayed as attributes different from other items.

Meanwhile, a search location display area 420 that displays a directory and a cluster that the user is searching may be additionally provided.

The cluster display area 430 is an area that provides a means of accessing documents belonging to a cluster that the user is searching. In the present embodiment, the central concept of the cluster that the user is searching for is 'Bordeaux' and documents related to the central concept of 'Bordeaux' may be provided in the cluster display area 430.

The cluster display area 430 is provided with a document link 434 to documents belonging to the cluster that the user is searching. The document link 434 may display title information of the referenced document as anchor text.

Meanwhile, the document link 434 may be grouped again around the overlapping portion of the anchor text. In this embodiment, document links 434 are listed one after the other after displaying the duplicate phrase 432.

In the present embodiment, the 'Bordeaux' cluster includes a group whose overlapping phrase is 'Nature of Bordeaux' and a group whose overlapping phrase is 'Bordeaux map'. The cluster may be composed of only one group, in which case, the redundant clause of the group may be used as the name of the cluster.

Document link 434 is a link to each document. By selecting this link, the user has access to the contents of the document containing the information he or she wants to search. In this case, the user terminal 230 may generate a new browser window to provide the content of the document, or may provide the content of the document through all or part of the browser window on which the document search service providing screen of FIG. 4 is displayed. Do.

The user can change his or her search target by selecting an item of each area provided in the present embodiment or the like. In response, the document providing unit 215 updates the classification structure display area 410, the search position display area 420, and the cluster display area 430.

Through the above screen configuration, the directory structure and the clustering structure can be visually transmitted to the user, thereby efficiently searching for information on the field of interest without separately visiting the vertical site and the blog that the user visits.

5 is a diagram illustrating a document search service providing screen including an advertisement display area according to an embodiment of the present invention. Referring to FIG. 5, a classification display area 400, a classification structure display area 410, a search position display area 420, a cluster display area 430, and an advertisement display area 510 are illustrated.

The advertisement display area 510 is an area where advertisement content provided by the advertisement providing unit 216 to the user terminal 230 is displayed. In the advertisement display area 510, text advertisement content 511 and animated advertisement content 512 are displayed.

The text advertisement content 511 and the animation advertisement content 512 may further include a link for accessing a site including additional information related to an advertiser.

The advertisement displayed in the advertisement display area 510 may include a pay per click (PPC) model in which the advertisement cost is executed in accordance with the number of clicks, and / or a PPV (PPV) in which the advertisement cost is executed in accordance with the number of impressions. pay per view) can be operated based on the model.

Meanwhile, as described above, information about the user and information about a document searched by the user may be considered as an element for determining the advertisement content displayed on the advertisement display area 510.

For example, information about a subject of a document, a directory, a cluster, and a title of a document may be considered as an advertisement content determining element. Referring to FIG. 5, the subject the user is searching for is 'wine', the directory is 'mountain of wine', and the cluster is 'Bordeaux'. Based on this information, it is possible to maximize the effect of the advertisement by providing the text advertisement content 511 titled 'Bordeaux wine co-purchase application'.

On the other hand, the document search service providing method can be created by a computer program. Codes and code segments constituting the program can be easily inferred by a computer programmer in the art. In addition, the program is stored in a computer readable media, and read and executed by a computer to implement a method for providing a document search service. The information storage medium includes a magnetic recording medium, an optical recording medium, and a carrier wave medium.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise.

In this application, the terms "comprise" or "having" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, one or more other It is to be understood that the present invention does not exclude the possibility of the presence or the addition of features, numbers, steps, operations, components, parts, or a combination thereof.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

So far I looked at the center of the present invention with respect to the embodiment. Many embodiments other than the above-described embodiments are within the scope of the claims of the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

1 is a flowchart of a method of providing a document search service according to an embodiment of the present invention.

2 is a block diagram of a system for providing a document search service according to an embodiment of the present invention.

3 is a diagram illustrating a cluster and directory mapping of a document according to an embodiment of the present invention.

4 is a diagram illustrating a document search service providing screen according to an embodiment of the present invention.

5 is a diagram illustrating a document search service providing screen including an advertisement display area according to an embodiment of the present invention.

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

210: search service providing server 211: document classification unit

212: title extracting unit 213: cluster forming unit

214: directory mapping unit 215: document providing unit

216: advertisement provider 221: original document database

222: navigation service database 223: advertising database

230: user terminal

400: classification display area 410: classification structure display area

420: navigation location display area 430: cluster display area

432: Duplicate verse 434: Document link

510: advertisement display area 511: text advertising content

512: animated advertising content

Claims (15)

Classifying documents according to subject matter; Extracting a title of the document; Forming a cluster based on the extracted title; and Mapping the cluster to a predetermined directory belonging to the subject. The method of claim 1, Providing the document to a user terminal; The document providing step provides a means for visualizing the structure of the directory and the cluster to which the document belongs. The method of claim 1, The document is characterized in that it comprises one or a plurality of fields, The extracting of the title of the document may include extracting the title in consideration of an attribute of a field configuring the document. The method of claim 1, Forming the cluster is And dividing the extracted title into syllable units and selecting a part of the title shared with other documents as a central concept candidate of the cluster. The method of claim 4, wherein Forming the cluster is And selecting the central concept candidate using n-gram analysis of the extracted title. The method of claim 4, wherein Forming the cluster is And a cluster is formed when there are more than a predetermined number of documents sharing the central concept candidate. The method of claim 4, wherein And a plurality of the central concept candidates, selecting the central concept in consideration of the length of the title candidate phrase. A computer-readable recording medium for recording a program for executing the method of any one of claims 1 to 7 on a computer. A document classification unit for classifying documents according to subjects; A title extractor for extracting a title of the document; A cluster forming unit forming a cluster based on the extracted title; and And a directory mapping unit for mapping the cluster to a predetermined directory belonging to the subject. The method of claim 9, Further comprising a document providing unit for providing the document to the user terminal, And the document providing unit provides a means for visualizing the directory to which the document belongs and the structure of the cluster. The method of claim 9, The document is characterized in that it comprises one or a plurality of fields, And the title extracting unit extracts the title in consideration of an attribute of a field constituting the document. The method of claim 9, The cluster forming unit And dividing the extracted title into syllable units, and selecting a part of the title shared with other documents as a central concept candidate of the cluster. The method of claim 12, The cluster forming unit The system for providing a document search service, characterized in that for selecting the central concept candidate using an n-gram analysis of the extracted title. The method of claim 12, The cluster forming unit And a cluster is formed when there are more than a predetermined number of documents sharing the central concept candidate. The method of claim 12, And a plurality of the central concept candidates, wherein the central concept is selected in consideration of the length of the title candidate phrase.

Images