JP2004013745A - Device and method for extracting document dependence - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system in which it can be considered that dependence exists between documents when a word defined in a certain document is used in a different document and which extracts dependence between a plurality of documents and supports document understanding. <P>SOLUTION: A defined attribute giving part 11 gives defined attribute to each document of a document group. For instance, a morpheme analysis is conducted so as to divide the document into words, and "defined attribute" is given to the word or the like which corresponds to "A" extracted by a language pattern such as "A is B". The defined attribute giving part 11 gives a word having the extracted defined attribute and an original document given to a reference attribute giving part 12. The reference attribute giving part 12 examines whether a word which is the same character string as the word having the defined attribute received and has no defined attribute exists in each of six original documents or not. When the word exists, the reference attribute is given to the word. A dependence link generating part 13 pasts a link to the word of the same character string having the defined attribute from the word having the reference attribute, quantitatively evaluates the link to generate the dependance between the documents and preserves it in a dependence link data base 15 by a dependence link preserving part 14. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a technique for extracting a dependency between documents.
[0002]
[Background technology]
When words defined in one document are used in another document, it can be considered that there is a dependency between the documents. It is desired to extract dependencies between a plurality of documents to support document understanding.
[0003]
That is, when a human reads a certain document, there may be a document serving as a premise for reading the document. For example, in order to read a certain part of a manual of a certain device, it is necessary to know terms such as a part name related to the device in advance.
[0004]
Also, when reading certain materials of a project, there is no problem if you are familiar with the project, but if you are not very familiar, if the terms etc defined in the project are used, you will not be able to read other materials. You need to look over it.
[0005]
Also, in classes at universities and the like, there may be cases where there are credits that should be learned in advance before the credits that the student wants to acquire. In other words, it may be better to read a teaching material document of another class before reading a teaching material document used in one class.
[0006]
In such a case, examine the dependencies between all the materials used in the project and all the teaching materials used in the university, and if you can recommend the documents to be read before reading a certain document, read the document if you can recommend it. I can help.
[0007]
The present invention has been made in response to the above needs.
[0008]
Prior art documents related to the present invention include the following.
(1) Japanese Unexamined Patent Publication No. 7-325827: "Automatic hypertext generator"
(2) JP-A-5-225247: "Method of displaying structure between documents"
(3) Japanese Patent Application Laid-Open No. 2000-259657: "Term Definition Search / Collection Device"
[0009]
In (1), the words included in the document are expanded into synonyms using matching or a thesaurus, and a link (hyperlink) is generated based on the matching between the character strings of the words, or the link source word is generated. Discloses that a link is generated to the title of a section in which the word frequently appears. Since it is a link between words, the dependency between documents cannot be extracted.
[0010]
In (2), the degree of relevance between documents is determined based on the number of common words included in two documents, and the graph is colored and the thickness of the link is set based on the value to determine the relation between the documents. It shows the strength of. If the same word is included, it is likely that the two documents are related, but it is difficult to judge whether to read or not because the relation is not known.
[0011]
In (3), a term and its definition part are extracted from a given document, and registered in a database so that a search can be performed. However, the background where the term came out, examples of the term, etc. are often described in the material in which the term is actually defined, and are defined rather than giving the user only the definition. Sometimes it is better to give a document or its page.
[0012]
[Problems to be solved by the invention]
The present invention has been made in view of the above circumstances, and has as its object to provide a technique for extracting a dependency between documents.
[0013]
[Means for Solving the Problems]
According to the present invention, in order to achieve the above-mentioned object, for example, for each word constituting each document included in a given document group, an attribute of whether the word is defined or referenced (used) is set. Given as "definition attribute" and "reference attribute", a link is established from a word having the reference attribute to a word of the same character string having the definition attribute, and the degree of dependence is calculated based on the link.
[0014]
That is, according to one aspect of the present invention, in order to achieve the above-described object, a document dependency extracting apparatus includes: a document storage for storing a plurality of document units each of which can be treated as a single document and distinguished from other documents; Means; first discriminating means for discriminating that a word appears with a definition (being a definition attribute) in each document unit stored in the document storage means; and stored in the document storage means. A second discriminating unit for discriminating that a word determined to have appeared with a definition by the first discriminating unit appears without a definition (that is, a reference attribute) in each document unit; Based on the correspondence between the word that appeared with the word and the word corresponding to the word and without the definition, the document unit in which the word that appeared with the above definition was determined, and the Appearing words are to be provided with a dependency determination means for determining dependencies between document units is determined without definition.
[0015]
In this configuration, it is possible to determine whether there is a probability that there is a definition-reference relationship between words of the same character string appearing in different document positions, and to easily determine the dependency between document units based on this. it can.
[0016]
The dependency may be measured using a thesaurus or the like in consideration of the relationship between the “definition attribute” and the “reference attribute” with a synonym. "Word" broadly refers to a component of a sentence, and includes a compound word.
[0017]
The present invention can be realized not only as a device or a system but also as a method. In addition, it goes without saying that a part of such an invention can be configured as software. Also, it goes without saying that a software product used for causing a computer to execute such software is also included in the technical scope of the present invention.
[0018]
The above aspects of the present invention and other aspects of the present invention are set forth in the following claims, and will be described in detail below with reference to embodiments.
[0019]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described.
[0020]
[Document dependency extraction method]
First, the principle of the document dependency extraction method used in the embodiment will be described.
[0021]
In this embodiment, for each word constituting each document included in a given document group, attributes that are defined or referenced (used) are given as “definition attribute” and “reference attribute”. A link is formed from a word having a reference attribute to a word having the same character string having a definition attribute, and the dependency is calculated based on the link.
[0022]
As a method of assigning the definition attribute, words defined by using a language pattern such as “is a to” are extracted, and “definition attributes” can be assigned to them.
[0023]
Further, a definition attribute may be given by using document layout information. For example, if you can extract the title of a document or chapter, and it is only words or compound words, it is likely that the document or section explains the word and it is defined as " Definition attribute ". If there is an item name such as "item name text" and there is a layout in which the explanation is continued, the definition attribute may be given assuming that the item name is explained.
[0024]
The reference attribute may be assigned to a word having the same character string as the word to which the above-described definition attribute has been assigned, but not having the definition attribute, to which the “reference attribute” has been assigned. If a word having a reference attribute has the same character string having a definition attribute, a link is provided from the word having the reference attribute to a word having the same character string having the definition attribute.
[0025]
When the above-mentioned link from a word of a certain document to a word of another document is attached, the degree of dependence between documents is represented based on the number of links and the like. If there is a dependency between certain documents, a dependent link having the dependency as an attribute value is placed between the documents. The dependency between the documents is visually shown to support the understanding of the documents.
[0026]
Dependency links are given a degree of dependency, so that when reading a certain document, only documents that depend on that document and have a certain degree of dependency or higher can be collected or ranked in descending order of dependency and presented to the user. .
[0027]
Dependency Dep (Da, Db) from document Da to document Db is calculated by the following equation.
(Equation 1)
Dep (Da, Db) = Σw (Kab [n]) (a ≠ b)
[0028]
Here, Kab is a set of words having the same attribute as the word having the definition attribute in the document Db and having the reference attribute in the document Da.
Kab [n]: the word indicated by n in the word set Kab (the number of words included in 1 ≦ n ≦ Kab).
w (Kab [n]): weight of word Kab [n].
[0029]
w (Kab [n]) is calculated from the number of times that the word Kab [n] appears with the reference attribute in the document Da and the number of times that the word appears with the definition attribute in the document Db.
(Equation 2)
w (Kab [n]) = ref_tf (Da, Kab [n]) * ref_w + def_tf (Db, Kab [n]) * def_w
[0030]
ref_tf (Da, Kab [n]) is the number of appearances of Kab [n] having the reference attribute in the document Da.
def_tf (Db, Kab [n]) is the number of appearances of Kab [n] having the definition attribute in the document Db.
ref_w and def_w can be changed by weight. For example, it may be calculated in advance from a document group given idf (inverted document frequency) in order to take into account the generality of words, a dictionary, a news article, or the like. By adding weights as described below, it is possible to reduce the weight of words considered to be general terms.
[Equation 3]
ref_w (Kab [n]) = [idf obtained from one year of newspaper article of word Kab [n]]
[0031]
[Document Registration Device]
Next, the document registration device 100 used in this embodiment will be described. The document registration apparatus 100 receives a plurality of documents and generates and registers a degree of dependency between the documents. The document registration device 100 can be realized by, for example, a stand-alone personal computer, or can be realized by a server arranged on a network.
[0032]
FIG. 1 shows the configuration of a document registration device 100. In this figure, the document registration device 100 includes a document group reception unit 10, a definition attribute assignment unit 11, a reference attribute assignment unit 12, a dependency link generation unit 13, It includes a link storage unit 14 and a dependent link database 15.
[0033]
The document group receiving unit 10 receives a document group specified by the user. The designation method is, for example, all the documents stored in a certain directory, but is not limited to this. It is assumed that the document group receiving unit 10 has received seven documents as shown in FIG.
[0034]
The document group receiving unit 10 passes the received document group to the definition attribute providing unit 11. The definition attribute assigning unit 11 assigns a definition attribute to each document.
[0035]
First, morphological analysis is performed to divide the words. Words or compound words equivalent to "~" extracted by language patterns such as "is a-" or "-is called-", or words used as titles or item names And compound words are defined and given a "definition attribute".
(1) From the document 1, the “protocol” which is the title and the “protocol” corresponding to the “to” part of the “to which − is defined” are extracted.
(2) From the document 2, "Network" corresponding to the title "Network", "Network" corresponding to "-" of "consisting of-", and "OSI" corresponding to "-" of "there is-". Reference model "is extracted.
(3) From the document 3, "Internet" which is a part of "to" of "is one of-" is extracted.
(4) From the document 4, “Packet communication” as a title is extracted.
(5) From the document 5, “Packet” as a title is extracted.
(6) From the document 6, “TCP / IP” as a title is extracted.
(7) From the document 7, “media” which is the title of “media” and “to is-” is extracted.
[0036]
The definition attribute assigning unit 11 passes the word having the extracted definition attribute and the given original document to the reference attribute assigning unit 12. The information passed to the reference attribute assignment unit 12 is as shown in FIG. In FIG. 3, “* n” represents the number (n).
[0037]
The reference attribute assigning unit 12 checks whether or not a word having no definition attribute in the original seven documents is the same character string as the received word having the definition attribute. If so, a reference attribute is assigned to the word. The determination result of the reference attribute assigning unit 12 is as follows.
(2) “Protocol” is extracted from document 2.
(3) “Internet”, “network”, “TCP / IP”, “packet communication”, and “media” included in the title are extracted from document 3.
(4) “Packet” is extracted from document 4.
(5) “TCP / IP” is extracted from the document 5.
(6) From the document 6, "Internet", "protocol", "OSI reference model", and "packet" are extracted.
(7) None from document 7.
[0038]
The reference attribute assigning unit 12 assigns a reference attribute to the extracted word, and passes the word to the dependent link generating unit 13. The reference attributes of the words are as shown in FIG.
[0039]
The dependent link generation unit 13 generates a link from the received definition attribute and reference attribute. The link generation method is to link a word having a reference attribute to a word having the same character string having a definition attribute. Ignore if the link destination is a file that contains itself. If the same word having the definition attribute exists in a plurality of documents, a link is attached to each word. In this example, links are provided as shown in FIG. In FIG. 5, the link is represented by "->".
[0040]
After creating the links, the degree of dependency between the documents is calculated. Here, ref_w and def_w are each set to 1. If the value is 0, ignore it.
(1) Dependency from document 2 to document 1 (@w (Kab [n])) = (1 * 1 + 2 * 1) = 3
(2) Dependency from document 3 to document 2 = (1 * 1 + 1 * 1) = 2
(3) Dependency from document 3 to document 4 = (1 * 1 + 1 * 1) = 2
(4) Dependency from document 3 to document 6 = (1 * 1 + 1 * 1) = 2
(5) Dependency from document 3 to document 7 = (1 * 1 + 1 * 1) = 2
(6) Dependency from document 4 to document 5 = (1 * 1 + 1 * 1) = 2
(7) Dependency from document 5 to document 6 = (2 * 1 + 1 * 1) = 3
(8) Dependency from document 6 to document 1 = (2 * 1 + 1 * 1) = 3
(9) Dependency from document 6 to document 2 = (1 * 1 + 1 * 1) = 2
(10) Dependency from document 6 to document 3 = (1 * 1 + 1 * 1) = 2
(11) Dependency from document 6 to document 5 = (3 * 1 + 1 * 1) = 4
[0041]
The dependency between the documents is passed to the dependency link storage unit 14. The dependency link storage unit 14 stores the dependency between documents in the dependency link database 15. The information on the degree of dependence passed to the dependence link database 15 is as shown in FIG.
[0042]
As described above, the dependency between documents is extracted and stored by the document registration device 100.
[0043]
[Dependency presentation device]
Next, a dependency presenting apparatus 200 that presents a dependency between documents using the degree of dependency extracted and stored by the document registration apparatus 100 of FIG. 1 will be described. The dependency relationship presentation device 200 is also configured by a stand-alone personal computer or server. The document registration device 100 and the dependency relationship presentation device 200 may constitute one device / system.
[0044]
FIG. 7 shows the configuration of the dependency presenting apparatus 200 of this embodiment. In this figure, the dependency presenting apparatus 200 includes a document name receiving unit 20, a dependent link searching unit 21, a document relationship presenting unit 22, It is configured to include a link database 15 and the like. The dependency link database 15 is the dependency link database of FIG.
[0045]
The user instructs the document name accepting unit 20 of the document name of the document that the user must read (learn). Here, it is assumed that the document 6 is designated. The document name receiving unit 20 passes the document name to the dependent link search unit 21. The dependent link search unit 21 searches the dependent link database 15 for a document on which the document 6 depends, and passes the search result to the document relation presenting unit 22. The document relation presentation unit 22 sorts the results in descending order of the degree of dependence and presents them to the user. For example, it is presented as shown in FIG. Alternatively, as shown in FIG. 9, the relationship between the document 6 and another document linked to the document 6 may be visually shown by thickening the link line depending on the degree of dependency.
[0046]
The above-described document registration device 100 and the dependency relationship presentation device 200 can be configured by, for example, a server device 400 arranged on a network 300 as shown in FIG. The server device 400 can be configured by a web server, an application server, and the like. The dependency between documents can be registered in the dependency link database 15 in response to a request from the client device 500, and the dependency can be displayed in response to a presentation request. Further, a process of an application program for learning support or browsing a manual document or the like may use information of the dependent link database 15. For installation of the server device 300, a recording medium 401 storing a program corresponding to the document registration device 100 or the dependency relationship presentation device 200 is used. This program may be sent from outside via a network.
[0047]
As described above, according to this embodiment, when a user reads a certain document, the user can assist reading of the document by presenting the document that is a prerequisite for reading the document.
[0048]
It should be noted that the present invention is not limited to the above-described embodiment, and various changes can be made without departing from the gist of the present invention. For example, in order to restrict links, assign a group name to a group of documents and restrict links only to that group, or set personal settings, and reduce the dependence on documents read by individuals as they already understand It is possible. It can be set manually by the user or automatically depending on whether the document is opened or not.
[0049]
【The invention's effect】
As described above, according to the present invention, it is possible to easily support the understanding of a document by extracting the dependency between documents.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration example of a document registration device 100 according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating the operation of the document registration device 100 of FIG.
FIG. 3 is a diagram illustrating the operation of the document registration device 100 of FIG.
FIG. 4 is a diagram illustrating the operation of the document registration device 100 of FIG.
FIG. 5 is a diagram illustrating the operation of the document registration device 100 of FIG.
FIG. 6 is a diagram illustrating an operation of the document registration device 100 of FIG.
FIG. 7 is a block diagram illustrating a configuration example of a dependency relationship presentation device 200 according to the above embodiment.
8 is a diagram illustrating the operation of the dependency relationship presentation device 200 of FIG.
FIG. 9 is a diagram illustrating the operation of the dependency relationship presentation device 200 of FIG.
FIG. 10 is a diagram illustrating an example of mounting the document registration device 100 and the dependency relationship presentation device 200 using a server device.
[Explanation of symbols]
10 Document Group Receiving Unit 11 Definition Attribute Giving Unit 12 Reference Attribute Giving Unit 13 Dependent Link Generating Unit 14 Dependent Link Storage Unit 15 Dependent Link Database 20 Document Name Receiving Unit 21 Dependent Link Searching Unit 22 Document Relationship Presentation Unit 100 Dependent on Document Registration Device 200 Relationship presentation device 300 Network 400 Server device 401 Recording medium 500 Client device

Claims (9)

A document storage unit for storing a plurality of document units each of which can be treated as one lump document separately from other documents;
A first determination unit that determines that a word appears with a definition in each document unit stored in the document storage unit;
A second discriminator for discriminating that a word determined to have appeared with a definition by the first discriminator has appeared without a definition in each document unit stored in the document storage;
Based on the correspondence between the word that appeared with the definition and the word corresponding to the word and appeared without the definition, a document unit in which the word that appeared with the definition was determined, A dependency determining unit for determining a dependency with respect to a document unit in which a corresponding word that has appeared without a definition is determined. 2. The document dependency extracting apparatus according to claim 1, wherein the first determining means determines that the word appears with a definition based on a sentence pattern. 3. The document dependency extracting apparatus according to claim 2, wherein the pattern of the sentence is a pattern of "is a ...". 4. The document dependency extracting apparatus according to claim 1, wherein the first determination unit determines that a word appears with a definition based on a predetermined layout in the document unit. 5. The document dependency extracting apparatus according to claim 4, wherein the predetermined layout is a layout in which a title of a document or a chapter is a word or a compound word. 5. The document dependency extracting apparatus according to claim 4, wherein the predetermined layout is a layout having an item name and a description following the item name, such as "item name @ text". Dependency storing means for storing a dependency between document units extracted by the document dependency extracting apparatus according to any one of claims 1 to 6,
Means for specifying a document unit that the user is using for access;
Means for extracting a dependency related to the specified document unit by referring to the dependency storage means based on the specified document unit;
Means for displaying the taken-out dependency degree. A document storing step of storing a plurality of document units each of which can be treated as one lump document separately from other documents;
A first determination step of determining that a word has appeared with a definition in each document unit stored in the document storage step;
A second determination step of determining, in each document unit stored in the document storage step, that a word determined to have appeared with a definition in the first determination step has appeared without a definition;
On the basis of the correspondence between the word that appeared with the definition and the word that appeared without the definition corresponding to the word, the document unit in which the word that appeared with the above definition was determined, A dependency determining step of determining a dependency on a document unit in which a word that has appeared without a definition is determined. A document storing step of storing a plurality of document units each of which can be treated as one lump document separately from other documents;
A first determination step of determining that a word has appeared with a definition in each document unit stored in the document storage step;
A second determination step of determining, in each document unit stored in the document storage step, that a word determined to have appeared with a definition in the first determination step has appeared without a definition;
On the basis of the correspondence between the word that appeared with the definition and the word that appeared without the definition corresponding to the word, the document unit in which the word that appeared with the above definition was determined, A dependency determining step of determining a dependency on a document unit in which a word appearing without a corresponding definition has been determined.

Images