JP2007140603A - Early adapter extraction method and device and program and topic word prediction method and device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract a writer who is sensitive to a topic word, and to extract an information description document described by an early adapter, and to analyze and extract a topic word which may be a topic included in the latest information description document. <P>SOLUTION: An information description document in which a topic word is included is extracted from a plurality of information description documents, and the fixed number of information description documents whose preparation dates are old, written in an early stage, are extracted from the information description documents, and a writer as an early adapter relating to a topic word describing the information description document whose preparation date is old is extracted. Also, the information description documents for specifying the writer and the preparation date are collected, and the document prepared in the described period is extracted from among the documents of the document writer as the early adapter relating to a certain topic word from among the information description documents, and the topic word is extracted from the document. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an early adapter extraction method and apparatus and program, and a topic word prediction method and apparatus and program, and more particularly to an early adapter extraction method and apparatus and program for extracting an author who presents the latest topic word from an information description document. , And a topic word prediction method, apparatus, and program for extracting topic words.

As a technique for extracting topic words from a wide variety of collected documents, when a plurality of categorized documents with creation time information are input, the document is analyzed, and phrases appearing in the document and phrases included There is a technique in which information is aggregated and the degree of relevance to a category in which a phrase appears is calculated for each phrase using the phrase accompanying information of the phrase (see, for example, Patent Document 1).
JP 2005-13531 A

  In the conventional technique for extracting topic words, analysis is performed by a technique based on extracting words having a high frequency of appearance in the entire collected documents, and topic words are extracted. Furthermore, the accuracy of topic word extraction has been improved by devising the analysis method itself on the basis of extracting words with high appearance frequency.

  However, the conventional topic word extraction technique can extract the topic word that is the current topic, but cannot predict and extract the topic word that becomes the topic.

  The present invention has been made in view of the above points, and extracts an author (early adapter) sensitive to a topic word and enables extraction of an information description document described by the early adapter, which is included in recent information description documents. It is an object of the present invention to provide an early adapter extraction method and apparatus and program, and a topic word prediction method and apparatus and program capable of predicting a topic word by analyzing and extracting a topic word to be discussed.

The present invention (Claim 1) is an early adapter extraction method for extracting an author who is an early adapter related to a topic word from a plurality of information description documents that can specify an author and a creation date.
Topic word filtering means for extracting an information description document including a topic word from the plurality of input information description documents,
A creation date filtering step in which a creation date filtering means extracts a predetermined number of information description documents written earlier in the creation date from the information description document;
The early adapter extracting means performs an early adapter extracting step of extracting an author who is an early adapter related to a topic word describing an information description document having an older creation date obtained in the creation date filtering step.

  FIG. 1 is a diagram for explaining the principle of the present invention.

The present invention (Claim 2) is a topic word prediction method for predicting the latest topic word from a plurality of information description documents that can specify the author and the creation date,
A document collection step (step 1) in which the document collection unit collects an information description document capable of specifying the author and the creation date and stores the information description document in the document storage unit;
The early adapter document extraction means extracts documents created during a specified period from the document author's documents that are early adapters related to a topic word from the information description document stored in the document storage means, and stores the early adapter document. An early adapter document extraction step (step 2) to be stored in the means;
The topic word extraction means performs a topic word extraction step (step 3) for extracting a topic word from the document stored in the early adapter document storage means.

Further, according to the present invention (Claim 3), in the early adapter document extraction step (Step 2),
A topic word filtering step in which the topic word filtering means of the early adapter document extracting means extracts an information description document including a topic word from the information description document stored in the document storage means;
A creation date filtering step in which the creation date filtering means of the early adapter document extraction means extracts a predetermined number of information description documents written earlier in the creation date from the information description documents extracted in the topic word filtering step;
The early adapter description document extracting means of the early adapter document extracting means obtains the author's information description document which is an early adapter related to the topic word describing the information description document with the old creation date from the information description document extracted in the creation date filtering step. An early adapter description document extraction step to be extracted;
Period filtering step in which the period filtering means of the early adapter document extracting means extracts a document created within a certain period from the author's information description document which is an early adapter related to the topic word obtained in the early adapter description document extracting step. And do.

The present invention (Claim 4) is an early adapter extraction device for extracting an author who is an early adapter related to a topic word from a plurality of information description documents that can specify an author and a creation date.
A topic word filtering means for extracting an information description document including a topic word from a plurality of input information description documents;
A creation date filtering means for extracting a fixed number of information description documents written earlier from an information description document,
An early adapter extracting means for extracting an author who is an early adapter related to a topic word describing an information description document having an old creation date obtained by the creation date filtering means.

  FIG. 2 is a principle configuration diagram of the present invention.

The present invention (Claim 5) is a topic word prediction device that predicts the latest topic word from a plurality of information description documents that can specify the author and the creation date,
Document storage means 102 for storing the collected documents;
Early adapter document storage means 104 for storing a document author's document as an early adapter;
A document collection unit 101 that collects an information description document that can specify an author and creation date and stores the information description document in the document storage unit 102;
The document created during the specified period is extracted from the document author's document, which is an early adapter related to a topic word, from the information description document stored in the document storage unit 102, and stored in the early adapter document storage unit 104. Adapter document extraction means 103;
A topic word extraction unit 105 that extracts a topic word from a document stored in the early adapter document storage unit 104.

Further, according to the present invention (claim 6), the early adapter document extracting means 103
Topic word filtering means for extracting an information description document including a topic word from the information description document stored in the document storage means 102;
A creation date filtering means for extracting a certain number of information description documents written earlier in the creation date from the information description documents extracted by the topic word filtering means;
An early adapter description document extracting means for extracting an author's information description document that is an early adapter related to a topic word describing an information description document having an older creation date from the information description document extracted by the creation date filtering means;
Period filtering means for extracting a document created within a certain period from the author's information description document which is an early adapter related to the topic word obtained by the early adapter description document extracting means.

  The present invention (Claim 7) is an early adapter extraction program that causes a computer to execute the steps of Claim 1.

  The present invention (Claim 8) is a topic prediction program that causes a computer to execute the steps of Claim 2 or Claim 3.

  As described above, according to the present invention, the early adapter extraction device can extract the author who is sensitive to the topic word and the information description document of the author. Furthermore, recent information description documents of authors who are sensitive to topic words include these or topic words that become topics. By analyzing and extracting them, topic words can be predicted.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  In the present invention, not all of a wide variety of documents are analyzed, but an information description document in which an author and a creation date can be specified is targeted, and a topic word or buzzword already published in newspapers or various media. An information description document that describes an article related to one topic word in the attention word (topic word) and is the author of the information description document that picks up the topic word at an early stage , Extract the author (early adapter) sensitive to the topic word, extract the information description document created by multiple Web page authors sensitive to the topic word, and extract the topic word using the information description document as the object of analysis To do. In this way, by extracting information description documents created by a plurality of document creators who have previously noticed topic words in advance and described articles related to the topic words, and extracting topic words from these information description documents, A topic word to be a topic can be predicted.

  In this embodiment, for simplification of description, a blog is used as an information description document that can specify an author and a creation date.

  FIG. 3 is a configuration diagram of the latest topic word prediction device according to the embodiment of the present invention.

  The latest topic word prediction device 300 shown in FIG. 1 includes a blog collection unit 301 that collects blogs from blog sites on the Internet (corresponding to the document collection unit in claims), and a blog entry storage unit 302 that stores the collected entries (billing). An early adapter document extracting device 303 for extracting a blog entry described by a blogger sensitive to a topic word (corresponding to an author who is an early adapter of claims) from the collected blog entries. (Corresponding to the early adapter extracting means in claims), an early adapter entry storing section 304 (corresponding to the early adapter document storing means in claims) for searching for these blog entries, and a topic word extracting section for extracting topic words 305. Among these, the blog entry storage unit 302 and the early adapter blog entry storage unit 304 are storage media such as disk devices.

  In the latest topic word prediction device 300 of the present invention, first, blog entries are collected from a blog site on the Internet by the blog collection unit 301 and stored in the blog entry storage unit 302.

  Next, in the early adapter document extraction device 303, by setting a period from a plurality of blogger blogs created early among blog entries dealing with topic words from blog entries stored in the blog entry storage unit 302, Extract the latest blog entries. For example, by specifying “a blog entry created in the last three days”, the early adapter document extraction apparatus 303 selects a recently created entry among the blog entries of the early adapter that has noticed the topic word information at an early stage. It can be collected.

  Finally, a topic word is extracted by the topic word extraction unit 305. Since the blog entry of the above early adapter includes an entry including a topic word that has not been generally noticed, the latest topic word can be predicted.

  Next, the early adapter document extraction device 303 will be described in detail.

  FIG. 4 shows the configuration of the early adapter extracting apparatus in one embodiment of the present invention.

  The early adapter document extraction device 303 extracts a blog entry related to a topic word by performing a keyword search using a topic word that is already open to the public as a query regarding the blog entry stored in the blog entry storage unit 302. A word filter unit 401, a creation date filtering unit 402 that extracts a specified number of blog entries from the blog entries of search results from the earliest creation date. An early adapter blog extracting unit 403 for extraction and a period filtering unit 404 for extracting blogs of a specified period before the specified date from the blogs of the early adapter are configured.

  The early adapter blog entry storage unit 304 extracts, for example, the latest blog entry from the blogger blog focused on the topic word at an early stage by the early adapter document extraction device 303 having the above configuration.

  Next, the operation of each element of the latest topic word prediction device 300 will be described using FIG. 3, FIG. 4, and FIG. FIG. 5 is a flowchart of the operation of the latest topic word prediction device according to the embodiment of the present invention.

  The blog collection unit 301 collects blog entries from blog sites on the Internet (step 101) and stores them in the blog entry storage unit 302. At this time, date data is extracted from the blog text and stored. The date can be specified by using a tag in the blog text. Further, the blog entry is indexed and stored in the blog entry storage unit 302 so that a full-text search by keyword can be performed (step 102).

  The topic word filtering unit 401 uses the topic words, buzzwords, and after attention (hereinafter referred to as the following) that are already published in newspapers, televisions, and Internet websites for the blog entries stored in the blog entry storage unit 302. (Referred to as topic word) as a query (step 103).

  Let Gx be the set of entries obtained as a result. Gx is a document set of a plurality of blog entries. The creation date filtering unit 402 extracts K entries Dk (k = 1, 2,..., K) in order from the blog entries with the oldest creation date and time for the document set Gx (step 104).

  Each Dk, which is a blog entry, includes a topic word X, and since it is a blog entry in which a description relating to the topic word X exists, Dk is a blog entry described by a blogger sensitive to the topic word. The early adapter blog extraction unit 403 extracts bloggers Pn (n = 1, 2,..., N) that are N authors describing each of the K entries (step 105).

  All documents EPn of blog entries described by N bloggers Pn are collected (step 106), and a period filtering unit 404 sets a set for the latest entry R days from the EPn to Opn (OPn (EPn) is extracted (step 107).

  The period filtering unit 404 stores the extracted blog entry set OPn in the early adapter blog entry storage unit 304 as a database (step 108).

  The operations from step 103 to step 108 are performed by the early adapter blog extracting unit 303. Step 107 and step 108 are switched in order to store the blog of the early adapter in the early adapter blog entry storage unit 304, and then extract the R entry for a new entry update date, and extract the topic word extraction unit 305. May be sent to

  The topic word extraction unit 305 may extract frequent words in OPn as an analysis method for extracting topic words by analyzing the document set Opn. Morphological analysis is performed on the text of OPn, a word is extracted from the text, a noun is extracted from the obtained set of words based on the attribute of the morpheme (step 109), and the word having the highest appearance frequency is displayed as a topic word (step 109). Step 110).

  In step 109, it is conceivable to extract a word using a specific expression extraction technique instead of a morpheme. In step 110, a word suitable as a topic word may be extracted using an existing topic word extraction method such as using TF / IDF (Term Frequency Inverse Document Frequency), and this method is not particularly limited. .

  In addition, a configuration in which the blog entry collected from the blog entry collection unit 301 is directly processed by the early adapter document extraction device 303 without being stored in the blog entry storage unit 302 can be easily considered.

  Note that the present invention can also be used for the purpose of extracting an author who is an early adapter. In this case, the early adapter blog extracting unit 403 can perform the processing up to step 105 described above. In this case, the processing of outputting the blog entry of the early adapter after step 106 and the process of extracting the recently created blog entry are unnecessary.

  Further, in the present invention, the processing from step 101 to step 105 and the processing from step 101 to step 110 are constructed as a program and installed in a computer used as an early adapter extraction device and a topic word prediction device. And can be distributed through a network.

  Further, the constructed program can be stored in a portable storage medium such as a hard disk device or a flexible disk / CD-ROM, and installed in a computer to be executed.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made within the scope of the claims.

  According to the present invention, some search portal sites provide a service that presents a link to a keyword search result for a document on the Internet using a topic word as an anchor text. In general, the topic words presented by such services are manually provided by experts who have paid attention to various media such as newspapers and television, and such services can be provided automatically.

It is a figure for demonstrating the principle of this invention. It is a principle block diagram of this invention. It is a block diagram of the newest topic word prediction apparatus in one embodiment of this invention. It is a block diagram of the early adapter extraction apparatus in one embodiment of this invention. It is a flowchart of operation | movement of the newest topic prediction apparatus in one embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Latest topic word prediction apparatus 101 Document collection means 102 Document storage means 103 Early adapter extraction means 104 Early adapter document storage means 105 Topic word extraction means 300 Latest topic word prediction apparatus 301 Blog collection part 302 Blog entry storage part 303 Early adapter document extraction Device 304 Early adapter blog entry storage unit 305 Topic extraction unit 401 Topic word filtering unit 402 Creation date filtering unit 403 Early adapter blog extraction unit 404 Period filtering unit

Claims (8)

An early adapter extraction method for extracting an author who is an early adapter related to a topic word from a plurality of information description documents that can specify an author and a creation date,
Topic word filtering means for extracting an information description document including a topic word from the plurality of input information description documents,
A creation date filtering step in which a creation date filtering means extracts a predetermined number of information description documents written earlier in the creation date from the information description document;
An early adapter extracting step in which an early adapter extracting means extracts an author who is an early adapter related to a topic word describing an information description document having an old creation date obtained in the creation date filtering step;
The early adapter extraction method characterized by performing. A topic word prediction method for predicting the latest topic word from a plurality of information description documents that can specify an author and a creation date,
A document collection step in which the document collection means collects an information description document that can specify the author and creation date and stores the information description document in the document storage means;
An early adapter document extraction unit extracts a document created during a specified period from documents of a document author who is an early adapter related to a topic word from the information description document stored in the document storage unit, and the early adapter An early adapter document extraction step for storing in the document storage means;
A topic word extracting unit that extracts a topic word from the document stored in the early adapter document storage unit;
The topic word prediction method characterized by having. In the early adapter document extraction step,
A topic word filtering step in which the topic word filtering means of the early adapter document extraction means extracts an information description document including a topic word from the information description document stored in the document storage means;
Creation date filtering means for extracting creation date filtering means of the early adapter document extracting means for extracting a predetermined number of information description documents written earlier and earlier from the information description document extracted in the topic word filtering step. Steps,
The information of the author who is the early adapter related to the topic word describing the information description document with the old creation date from the information description document extracted in the creation date filtering step by the early adapter description document extraction means of the early adapter document extraction means An early adapter description document extraction step for extracting a description document;
Period filtering in which the period filtering means of the early adapter document extracting means extracts a document created within a certain period from the author's information description document which is an early adapter related to the topic word obtained in the early adapter document extracting step. Steps,
The topic word prediction method according to claim 2, wherein: An early adapter extraction device for extracting an author who is an early adapter related to a topic word from a plurality of information description documents that can specify an author and a creation date,
A topic word filtering means for extracting an information description document including a topic word from the plurality of input information description documents;
Creation date filtering means for extracting a predetermined number of information description documents written earlier in the creation date from the information description document;
Early adapter extraction means for extracting an author who is an early adapter related to a topic word describing an information description document with an old creation date obtained by the creation date filtering means;
An early adapter extraction device comprising: A topic word prediction device that predicts the latest topic word from a plurality of information description documents that can specify an author and a creation date,
Document storage means for storing the collected documents;
Early adapter document storage means for storing the document author's document as an early adapter;
A document collection unit that collects an information description document that can specify an author and a creation date, and stores the document in the document storage unit;
A document created during a specified period is extracted from documents of a document author who is an early adapter related to a topic word from the information description document stored in the document storage unit, and is stored in the early adapter document storage unit. Early adapter document extraction means;
Topic word extraction means for extracting a topic word from the document stored in the early adapter document storage means;
A topic word prediction device characterized by comprising: The early adapter document extraction means includes:
Topic word filtering means for extracting an information description document including a topic word from the information description document stored in the document storage means;
A creation date filtering means for extracting a fixed number of information description documents written earlier in the creation date from the information description document extracted by the topic word filtering means;
Early adapter description document extraction means for extracting an author's information description document that is an early adapter related to a topic word describing an information description document with an old creation date from the information description document extracted by the creation date filtering means;
Period filtering means for extracting a document created within a certain period from the author's information description document which is an early adapter related to the topic word obtained by the early adapter description document extracting means;
The topic word prediction device according to claim 5, comprising: On the computer,
An early adapter extraction program for executing the steps according to claim 1. On the computer,
A topic prediction program for executing the steps according to claim 2 or 3.

Images