JP2005222235A - Relevance presentation method, system and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a relevance presenting method for intelligibly presenting one or more optimal objects for each individual and group from among a plurality of objects. <P>SOLUTION: This relevance presenting method is provided with: an object input process for inputting a plurality of objects being the target of investigation; a word input process for inputting a plurality of words expressing the characteristics of each of the objects and the weight of each word in the object; an object vector generating process for generating an object vector expressing a relation between each of the plurality of objects and each word on the basis of the result of the word input process; a first vector calculating process for calculating a norm between arbitrary two object vectors among the object vectors; and a display process for displaying the relation between objects by using the norm. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a method and system for evaluating and presenting a relationship between a plurality of things, and a computer program therefor.

  The need to search for and present information that fits a particular purpose often arises. For example, search and classify papers that are closely related to technical papers in technical development work, present relations between educational subjects when providing education or taking classes, Searching for subjects to take.

  In the technical paper search example, a group of words (hereinafter referred to as keywords) expressing the characteristics of individual papers is given to the papers to be searched in advance, and information matching the search words and keywords is presented. In addition, a method has been adopted in which a word associated with a search word or a keyword is also used for determining a match. There has also been a method in which full-text search is performed to determine the appearance frequency of a search word, and a high-frequency one is preferentially presented (see Patent Documents 1 to 5).

In the educational situation, the education provider can edit each subject in advance into the educational system (curriculum) and let the students take the course, or in the case of the educational industry for working adults, education that can list the educational subjects A guide was provided to the student side, and the student selected from among them. Even here, there was a case where the subject search by keyword was performed.
JP 2002-15001 A JP 2003-157262 A Japanese Patent Laid-Open No. 2003-16089 JP-A-8-55136 JP 2002-41557 A

  However, the necessity of education is generally different for each individual and each group. In order to solve this problem, the side providing education, for example, divides education subjects into compulsory subjects and elective subjects, and the latter tries to solve this different necessity. However, if the range of selection is wide, for example, if it is possible to select from tens or hundreds of subjects, it will be extremely difficult for the student to make an effective selection. Especially in the case of working adults, if there is a waste of expenses and time related to education, this will immediately worsen corporate profits and losses, reduce the time to engage in the work, or in the case of self-development, Since problems such as an increase in cost burden occur, it is necessary to avoid such waste as much as possible.

  The necessity of selectively acquiring the optimum information in this way is not limited to the selection scene of the subject subject. For example, it is common in scenes such as searching for technical papers that are useful for business, or searching for color or color samples such as clothing and cosmetics that suit customer preferences. In this specification, search objects such as educational subjects, technical papers, color samples, and the like are represented in terms of things.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a relationship presentation method or system for presenting one or a plurality of optimal items for each individual or group in an easy-to-understand manner from among a plurality of items.

  The present invention achieves the above-mentioned object, and the invention according to claim 1 includes a thing input step for inputting a plurality of things to be investigated, a plurality of words expressing the characteristics of each thing, and A word input step for inputting the weight of each word in the thing, and an event vector generation step for generating a thing vector expressing a relationship with each word based on a result of the word input step for each of the plurality of things; A first vector calculation step of calculating a norm between any two of the event vectors, and a display step of displaying a relationship between the events by the norm. This is a relationship presentation method.

  Further, the invention according to claim 2 is the relationship presentation method according to claim 1, wherein the norm is based on a relationship between an object specifying step for specifying at least one of the things and the specified thing. And a sorting step for sorting things other than the specified thing based on the magnitude relationship of, wherein the display step displays a result of the sorting step.

  The invention according to claim 3 is the relationship presentation method according to claim 2, wherein the thing input step, the word input step, the matter vector generation step, the vector calculation step and the sort step are executed by a server side computer. The thing specifying step and the display step are performed using a client side computer connected to the server side computer via a communication line.

  The invention according to claim 4 is the relationship presentation method according to any one of claims 1 to 3, wherein the display step displays a combination of the two of the things in two dimensions, It includes a three-dimensional display that displays the norm corresponding to the combination in a third dimension.

  The invention according to claim 5 is the relationship presentation method according to claim 1, wherein a request vector creating step of creating a request vector by inputting a weight for each word, the request vector and the plurality of things A second vector calculation step of calculating a norm between each of the vectors, and a sorting step of sorting the things based on a norm magnitude relationship between the desired vector and each thing vector. The display step displays the result of the sorting step.

  The invention according to claim 6 further includes a display number input step of inputting a display output number in the relationship presentation method according to claim 2 or 5, wherein the display step is a matter of the display output number. Only the display.

  The invention according to claim 7 further includes a display limit norm input step of inputting a display limit norm in the relationship presentation method according to claim 2 or 5, wherein the display step includes the display of the norm as the display. It is characterized by only displaying things below the norm limit.

  The invention according to claim 8 is the relationship presentation method according to claim 5, wherein the thing input step, the word input step, the matter vector generation step, the vector calculation step, and the sort step are executed by a server-side computer. The thing specifying step, the request vector creating step, and the display step are performed using a client side computer connected to the server side computer via a communication line.

  The invention according to claim 9 is the relationship presentation system including at least one server-side computer and a plurality of client-side computers connected to the server-side computer via a communication line. , A thing input means for inputting a plurality of things to be investigated, a plurality of words expressing characteristics of each thing, a word input means for inputting the weight of each word in the thing, and each of the plurality of things A thing vector generating means for generating a thing vector expressing a relationship with each word based on the result of the word input, and a norm between two arbitrary thing vectors among the thing vectors. No. 1 is calculated by the relationship between the vector calculation means 1 and the thing specified by the client side computer. Sorting means for sorting things other than the specified thing based on the magnitude relationship of the above, the client side computer, the thing specifying means for specifying at least one of the things, and the norm And display means for displaying the relationship between the things based on the result of the sorting means.

  The invention according to claim 10 is the relationship presentation system according to claim 9, wherein the client side computer further includes a request vector creating means for creating a request vector by inputting a weight for each word. The server-side computer further includes second vector computing means for computing a norm between the desired vector and each of the plurality of thing vectors, and the display means displays a result of the sorting means. It is characterized by.

  The invention according to claim 11 is a computer for inputting a thing input procedure for inputting a plurality of things to be investigated, a plurality of words expressing characteristics of each of the things, and a weight of each word in the thing. A word input procedure for generating a thing vector representing a relationship with each word based on a result of the word input procedure for each of the plurality of things, and an arbitrary one of the thing vectors A relationship presentation program for executing a first vector calculation procedure for calculating a norm between two thing vectors and a display procedure for displaying a relationship between the things by the norm.

  The invention according to claim 12 is the relationship presentation program according to claim 11, wherein a request vector creation procedure for creating a request vector by inputting a weight for each word, the request vector and the plurality of things A second vector calculation procedure for calculating a norm between the vectors, and a sorting procedure for sorting the things based on a norm magnitude relationship between the desired vector and each thing vector. The program is characterized in that the display procedure displays the result of the sorting procedure.

  According to the present invention, for each individual or group, an object such as an educational subject or a technical paper, and a group of things can be visually provided with quantitative priorities and relationships between the objects. For this reason, it can be used as a decision support for selectively acquiring things.

  Hereinafter, an embodiment of the present invention will be described again by taking an educational subject as an example. FIG. 1 is a schematic view of an embodiment of the present invention. First, a word group (keyword) expressing the characteristics of a subject is acquired (step 1), and the relationship between the subject and each word group is weighted. For example, the weighting may be performed in the classification of (1) strong relationship: 3, (2) medium relationship: 2, (3) a little relationship: 1, and (4) a slight relationship: 0.

  Next, a subject vector S having these weights as components for each subject is defined (step 2). The plurality of subject vectors obtained in this way becomes a vector space model database.

  Next, arbitrary two subject vectors S and S ′ are selected from the database, and their norms (distance | S−S ′ |) are numerically calculated (step 3). At this time, the norm of each subject pair when the maximum norm found for all subject pairs is normalized to 1 is defined as subject distantness, and subject proximity is calculated by the equation of subject proximity = 1-subject remoteness. Numerical calculation may be performed by defining the degree.

  The technical reason for using the subject proximity defined in this way is as follows. The similarity between vectors can also be calculated by the cosine of both vectors. When the value is 1, both vectors are most similar. However, in this case, the norm of the vector will be ignored, and the difference in similarity strength itself cannot be expressed. In addition, correlation coefficient can be considered as a statistic for collecting information, but the correlation coefficient indicates the degree of increase / decrease change of the same component in the same phase, opposite phase, or random for two subject vectors. It is an indicator. That is, the correlation coefficient expresses only the relationship of the mutual direction with respect to the average deviation vector obtained by subtracting the average value from each element of the subject vector. On the other hand, the degree of subject proximity is an index that represents the similarity between two subjects, including the difference in intensity, so that the attendee can perform more effective subject selection based on the indicator.

  Finally, the subject group and subject proximity are visualized and displayed three-dimensionally (step 4).

  A first embodiment of the present invention will be described with reference to FIGS. FIG. 2 is an explanatory diagram of the function sharing of the client / server system in this embodiment. For example, the server 10 is installed on the side of the education providing entity (department, company, etc.), and there are a plurality of clients 12 corresponding to the side receiving the education (the student side). The server 10 and the client 12 can communicate bidirectionally via a communication network.

  The server 10 includes a subject vector creation unit 14, a subject vector database 24, a vector calculation unit 26, a norm database 28, a set value storage unit 30, and a sorting unit 32.

  The subject vector creation means 14 inputs the implementation subject group 18 via the subject group acquisition means 16 and the word group 22 via the word group acquisition means 20. With these inputs, the subject vector creation means 14 creates the subject vector database 24. Next, the vector calculation means 26 calculates the norm (distance) of any two vectors in the subject vector database 24, and the calculation result is stored in the norm database 28.

  The client 12 includes a selected subject setting unit 36, a predetermined output number setting unit 38, and an output unit 40. For example, the selected subject 42 that the student at the client 12 side felt interesting after attending the previous session is input by the selected subject setting means 36. Further, the output number 44 is set by the predetermined output number setting means 38.

  The selected subject 42 and the output number 44 transmitted from the client 12 side are stored in the set value storage means 30 on the server 10 side.

  The sorting means 32 on the server 10 side sorts the norm data in the norm database 28 according to the setting from the client. This sort result is output to the output means 40 on the client 12 side.

  FIG. 3 is an example of the subject vector database 24. The word group 22 is represented in a row, the subject group 18 is represented in a column, and weights (0 to 3) are given to the intersecting columns. To acquire the word group 22, it is collected and edited from subject planners / planners and lecturers so as to maintain the harmony between the completeness and the specificity that are mutually exclusive, and the weight is also collected at the same time. For example, the vector display of the subject “vision creation” is given by (2, 3, 2, 0, 0, 0, 0, 0, 0, 0, 2, 2, 1, 1, 0, 3, 1). . The vector was “17 dimensions”, and the number of vectors (number of subjects) was six. The setting from the client was “vision creation” for the standard subject and five outputs.

  FIG. 4 shows a display example by the subject proximity display means 4 (FIG. 1). The subject group 18 is arranged in two-dimensional coordinates, and the subject proximity is displayed in a three-dimensional manner with the height of the bar at the intersection. The bars at the intersections on the diagonal are not displayed, but the proximity between the same subjects is at those positions, so the height of the bar at that position is 1.

  According to the first embodiment, the degree of subject proximity between all subjects is displayed visually, and the similarity between subjects is easy to see. There is an effect that can be selected.

  Next, a second embodiment of the present invention will be described with reference to FIGS. This is an embodiment in which a recommended course menu is provided according to the objective of each student or group.

  FIG. 5 is an explanatory diagram of the function sharing of the client / server system in the embodiment. The difference from the first embodiment is that a request vector creating means 50 is added to the client 12. The request vector creating means 50 creates a request vector 52 representing a demand on the attending side. This request vector 52 is vector-calculated by the vector computing unit 26 together with the subject vector database 24 of the server 10.

  FIG. 6 is a table showing the request vector 52 added to the subject vector database 24 of FIG. Like the subject vector, the demand vector represents the degree of demand on the student side with respect to each element of the word group 22 by 0-3. In the example of FIG. 6, the request vector 52 is given by (2, 0, 2, 3, 2, 1, 2, 1, 0, 0, 2, 2, 1, 1, 0, 0, 2).

  FIG. 7 is a display example by the subject proximity display means 4 (FIG. 1). In this example, the setting from the client 12 is a request vector as a reference subject, that is, the number of outputs of recommended courses recommended to be close to “request from the student” is 4. According to the second embodiment, since the recommended course menus are shown in order, the attendees can select and take customized courses according to their own desire to study. There is an effect.

  Note that the method described in the above-described embodiment can be applied to various apparatuses by being written in a storage medium such as a magnetic disk, an optical disk, or a semiconductor memory as a program that can be executed by a computer, or transmitted by a communication medium. It can also be applied to various devices. A computer that implements the present system reads the program recorded in the storage medium, and performs the above-described processing by controlling the operation by this program.

The schematic flowchart of 1st Embodiment of the relationship presentation method which concerns on this invention. FIG. 3 is a block diagram showing the function sharing and the flow of operation of the client / server system in the first embodiment. A table showing an example of a subject vector database. The figure which shows the example of a three-dimensional display showing the proximity of subjects according to 1st Embodiment. The block diagram which shows the function sharing of the client server system in 2nd Embodiment, and the flow of operation | movement. An example of a table when a request vector is added to the subject vector database. The figure which shows the example of a three-dimensional display showing the proximity | contact degree of the subject mutual and the desired vector by 2nd Embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Server, 12 ... Client, 14 ... Subject vector creation means, 16 ... Subject group acquisition means, 18 ... Implementation subject group, 20 ... Word group acquisition means, 22 ... Word group, 24 ... Subject vector database, 26 ... Vector calculation Means 28 ... Norm database 30 ... Setting value storage means 32 ... Sort means 36 ... Selected subject setting means 38 ... Predetermined output number setting means 40 ... Output means 42 ... Selected subject 44 ... Number of outputs 50 ... request vector creation means, 52 ... request vector.

Claims (12)

An object input process for inputting a plurality of objects to be investigated;
A word input step of inputting a plurality of words expressing the characteristics of each thing and the weight of each word in the thing;
For each of the plurality of things, an event vector generation step for generating an event vector expressing a relationship with each word based on the result of the word input step;
A first vector calculation step of calculating a norm between any two of the event vectors;
A display step of displaying a relationship between the things by the norm;
The relationship presentation method characterized by having. The relationship presentation method according to claim 1,
An object specifying step for specifying at least one of the things;
A sorting step of sorting things other than the specified thing based on the magnitude relationship of the norm in relation to the specified thing;
Further comprising
The display step is to display a result of the sorting step;
A relationship presentation method characterized by The relationship presentation method according to claim 2,
The thing input step, the word input step, the thing vector generation step, the vector calculation step and the sort step are executed by the server side computer,
The thing specifying step and the display step are performed using a client side computer connected to the server side computer via a communication line;
A relationship presentation method characterized by In the relationship presentation method according to any one of claims 1 to 3,
The display step includes a three-dimensional display that displays a combination of two of the things in two dimensions and displays the norm corresponding to each combination in a third dimension;
A relationship presentation method characterized by The relationship presentation method according to claim 1,
A request vector creation step of creating a request vector by inputting a weight for each word;
A second vector calculation step of calculating a norm between the desired vector and each of the plurality of thing vectors;
A sorting step of sorting the things based on a norm magnitude relationship between the request vector and each thing vector;
Further comprising
The display step is to display a result of the sorting step;
A relationship presentation method characterized by In the relationship presentation method according to claim 2 or 5,
A display number input step for inputting the display output number;
The display step is to display only the items of the display output number;
A relationship presentation method characterized by In the relationship presentation method according to claim 2 or 5,
A display limit norm input step for inputting a display limit norm;
The display step is to display only the thing whose norm is not more than the display limit norm;
A relationship presentation method characterized by The relationship presentation method according to claim 5,
The thing input step, the word input step, the thing vector generation step, the vector calculation step and the sort step are executed by the server side computer,
The thing identifying step, the request vector creating step and the displaying step are performed using a client side computer connected to the server side computer via a communication line;
A relationship presentation method characterized by In a relationship presentation system having at least one server-side computer and a plurality of client-side computers connected to the server-side computer via a communication line,
The server computer is
An object input means for inputting a plurality of objects to be investigated;
A word input means for inputting a plurality of words expressing the characteristics of each thing and the weight of each word in the thing;
For each of the plurality of things, an event vector generation means for generating an event vector expressing a relationship with each word based on the result of the word input;
First vector computing means for computing a norm between any two of the thing vectors.
Sorting means for sorting things other than the specified thing based on the magnitude relation of the norm in relation to the thing specified by the client side computer;
Have
The client side computer
An item specifying means for specifying at least one of the items;
Display means for displaying a relationship between the things by the norm based on a result of the sorting means;
A relationship presentation system characterized by comprising: In the relationship presentation system according to claim 9,
The client side computer further has a request vector creating means for creating a request vector by inputting a weight for each word,
The server-side computer further includes second vector calculation means for calculating a norm between the desired vector and each of the plurality of thing vectors,
The display means is for displaying a result of the sorting means;
A relationship presentation system characterized by On the computer,
A matter input procedure for inputting a plurality of matters to be investigated,
A word input procedure for inputting a plurality of words expressing the characteristics of each of the things and the weight of each word in the things;
For each of the plurality of things, an event vector generation procedure for generating an event vector expressing a relationship with each word based on the result of the word input procedure;
A first vector calculation procedure for calculating a norm between any two of the event vectors;
A display procedure for displaying the relationship between the things by the norm;
A relationship presentation program to execute. In the relationship presentation program according to claim 11,
A request vector creation procedure for creating a request vector by inputting a weight for each word;
A second vector calculation procedure for calculating a norm between the desired vector and each of the plurality of thing vectors;
A sorting procedure for sorting the things based on a norm magnitude relationship between the request vector and each thing vector;
Is a program that further executes
The display procedure is to display a result of the sorting procedure;
A relationship presentation program characterized by

Images