JP2011181109A - Information retrieval support program, computer having information retrieval support function, server computer and program storage medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract an appropriate word as a retrieval keyword from a document described by a language which is not written with a space between words. <P>SOLUTION: An information retrieval support program executes steps of: managing each word in a document by associating grammatical information with each word; recognizing each character string unit of words, clauses and sentences, based on the grammatical information; dividing the document by each character string unit specified by a user and displaying a described part of the character string unit by making it identifiable as a part to be linked; setting character strings included in the part to be linked as the retrieval keyword; setting hyperlink that the retrieval keyword is referred to a predetermined database to the part to be linked; and displaying and outputting reference results returned from the predetermined database, based on the hyperlink, when the hyperlink set a certain part to be linked is instructed by the user in a computer. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  This invention applies morphological analysis technology, and automatically extracts search keywords for users to query various databases such as search engines and dictionary databases from documents that can be acquired as electronic data, or exists on the Internet or in document databases The present invention relates to a computer having a function for supporting work when searching for document information to be performed, a program installed in the computer to realize these functions, and a storage medium for the program.

  With the widespread use of general-purpose computers such as personal computers and information communication networks such as the Internet, many documents have been viewed on the computer display. And if the user wants to understand the document they are browsing more deeply, the words contained in the document can be used as search keywords, using well-known search engines, dictionaries / encyclopedia sites on the Internet, By giving it to a database on an appropriate network, information related to the word and the meaning of the word can be acquired instantly.

  Here, as a specific example, a case where a search keyword is given to a search engine is given. When a user is browsing a document displayed on a display by using document creation software installed on the computer, the same computer is used. A browser installed in is started, and the computer is made to access a Web site called a search site.

  A computer system that realizes a search site includes a WWW server and a search engine. The WWW server serves as an interface with the browser, and a web page (search) serving as a window for an information search service in the browser of the accessing user's computer. Return the contact page).

  A user inputs an appropriate word in the document as a search keyword by, for example, transferring it to a search keyword input field arranged on the Web page of the window by, for example, well-known “copy and paste”. When the search execution button on the same Web page is clicked with the mouse, the browser sends the search keyword to the WWW server.

  The WWW server gives the search keyword sent from the browser to the search engine. Search engines store and manage information (records), which are clues to information that users ultimately want to obtain, such as the contents and location of vast information on the Internet, in an accompanying database. The search keyword received via the WWW server is referred to the database, and the corresponding record is returned to the WWW server. As a search result, the WWW server creates a list of records received from the search engine on a Web page and returns it to the browser. When the user finds a noteworthy record in the search result list, the user designates a hyperlink set in the record. The browser obtains and displays information such as the Web page that originated from the record.

  In this way, users can obtain information related to the words in the document while browsing the document, deepen their understanding of the contents of the document, and obtain a wide variety of information related to the document. it can.

  In the above specific example, when a word included in a document being viewed on a computer is given to a search engine as a search keyword, (i) the browser is started, (ii) a search window page is obtained, and (iii) the search window page For those who are unfamiliar with computer operations, it is necessary to perform a series of tasks such as entering the words in the document in the keyword input field and (iv) instructing the sending of the entered search keywords. is there.

  Certainly, it may be possible to automatically set a hyperlink to give the word to the search engine at the place where the word is included in the document, but the document written in a language such as Japanese is English. Like a document written in, it is not separated into words by spaces. Moreover, there are various parts of speech in words, and particles that do not make sense as search keywords are also included. In addition, words with different parts of speech are described consecutively in a document that is not divided. For example, the character string “delicious meal” in the document can be broken down into two words: the adjective “delicious” and the noun “meal”. However, in order to obtain the target information, the user actually uses only the word “delicious” or the phrase “delicious meal” rather than performing an AND search with two words “delicious” and “meal”. It may be more appropriate to search. As described above, it is extremely difficult to appropriately select a search keyword not only for those who are not familiar with computer operation but also for those who are not familiar with information search technology.

  There is also a problem with the procedure itself for retrieving information by indicating a hyperlink. For example, as in the specific example described above, even if information that is the origin of a record is obtained by instructing a hyperlink, the information is not necessarily useful information for the user. That is, it is necessary to instruct a hyperlink and actually browse the linked page, which takes a lot of time and labor for information retrieval.

  An object of the present invention is to extract an appropriate word as a search keyword from a document described in a non-separated language, perform an information search based on the word, and present a search result to a user, or a hyperlink It is an object of the present invention to provide a computer having a function for supporting a user's information search operation by reducing time and labor when acquiring information while tracing the information. Another object of the present invention is to provide an information search support program that is installed in a computer and causes the computer to realize the information search support function, and a program storage medium thereof.

A first invention for achieving the above object is a program installed in a computer, the computer comprising:
Accepts user input indicating any character description position of the acquired document, extracts characters before and after the character at the character description position in character string units based on punctuation marks or reading marks, and describes the extracted character string A keyword extraction range setting step for setting a location as a keyword extraction range;
Morphological analysis of the character string of the keyword extraction range as a processing target, decomposed in units of words, and for each decomposed word, morpheme analysis step that associates and manages each part of speech and usage as grammatical information;
A character string unit recognition step for recognizing each character string unit of a word, a phrase, and a sentence from the character string to be processed based on the grammatical information;
In accordance with user input designating a character string unit, a keyword setting step for classifying the character string to be processed for each designated character string unit, and setting a character string described by the divided character string unit as a search keyword;
A keyword inquiry step of accepting a predetermined user input, inquiring the set search keyword to a predetermined database, and displaying and outputting an inquiry result returned from the database;
Is an information retrieval support program for executing

  According to a second aspect of the present invention, in the information search support program according to the first aspect, in the keyword extraction range setting step, the keyword extraction range is set according to the number of times a predetermined user input is continuously received within a predetermined time. It was decided to set the character string range.

A third invention is a program installed in a computer, and the computer includes
A morphological analysis step for analyzing a morphological analysis of a document acquired as a processing target and disassembling it into word units, and associating and managing each part of speech and usage as grammatical information for each decomposed word;
A character string unit recognition step for recognizing each character string unit of a word, a phrase, and a sentence from a continuous character string in a document based on the grammatical information;
A link target setting step of classifying the document into character string units designated by user input, and displaying the segmented character string unit so that it can be identified as a link target location;
A keyword setting step of setting a character string included in the link target portion as a search keyword;
A link setting step for setting a hyperlink indicating that the search keyword is referred to a predetermined database in the link target portion;
When a hyperlink set in a certain link target location is instructed by user input, a query result presentation step for displaying and outputting a query result returned from the predetermined database based on the hyperlink;
This is an information retrieval support program that executes

A fourth invention includes a syntax analysis step in the third invention, wherein the document is parsed to extract a sentence, and a dependency relationship in the sentence is acquired and managed.
In the link target setting step, in the sentence, when the preceding and following clauses are in a dependency relationship, the description location of the character string connecting the clauses in the dependency relationship is set as the link target location,
In the keyword setting step, an information search support program for setting a character string having a dependency relation designated by a user input in the sentence as a search keyword.

  In any one of the first to fourth inventions, in the keyword setting step, an information search support program that sets a character string obtained by converting a prescriptive word into a final form as a search keyword is a fifth invention. .

  According to a sixth invention, in any one of the first to fifth inventions, a word-specific appearance frequency obtaining step for obtaining an appearance frequency of each word included in the document, and a predetermined number of words in descending order of appearance frequency by user input. The information retrieval support program includes a keyword query step for querying a database as a search keyword.

  A seventh invention is an information search support program including a keyword part-of-speech setting step for setting a specific part-of-speech as the search keyword by a user input in any one of the first to fifth inventions.

  In any one of the first to seventh inventions, in the keyword setting step, an information search support program for setting a composite noun as a character string composed of consecutive nouns and setting the composite noun as one search keyword is It was set as the invention.

  A ninth invention is the information search support program according to the eighth invention, wherein, in the keyword setting step, the compound nouns are set by continuing the nouns in a number designated by a user input.

The present invention extends to a computer having an information search support function for extracting a search keyword to be referred to a predetermined database from an acquired document. The tenth invention is a client computer in a server-client system,
Means for displaying the acquired document;
Means for communicating with a server computer for executing morphological analysis processing;
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting means for extracting a character string unit and setting a description location of the extracted character string as a keyword extraction range;
An analysis request message sending means for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
When a message containing a search keyword is returned from the server computer that has received the analysis request message, search result presenting means for querying the search keyword to a predetermined database and displaying the query result is provided. ing.

A client computer in a server client system,
Means for displaying the acquired document;
Means for communicating with a server computer for executing morphological analysis processing;
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting means for extracting a character string unit and setting a description location of the extracted character string as a keyword extraction range;
An analysis request message sending means for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
Search result presentation means for displaying and outputting a search keyword query result returned from the server computer that has received the analysis request message;
The computer having the above is defined as an eleventh invention.

  In a twelfth aspect based on the tenth or eleventh aspect, the keyword extraction range setting means sets a character string range of the keyword extraction range according to the number of times a predetermined user input is continuously received within a predetermined time. Computer.

A thirteenth invention is a server computer that communicates the computer according to the tenth invention as a client computer,
The character string contained in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and for each decomposed word, each part of speech and usage form are associated with each other as grammatical information and managed. Morphological analysis means to perform,
Character string unit recognition means for recognizing each character string unit of words, phrases, and sentences from the character string based on the grammatical information;
Keyword setting means for dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
Search keyword inquiry message sending means for returning to the client computer a keyword inquiry message including the search keyword and an instruction for making the keyword inquired of a predetermined database.

14th invention is a server computer which communicates the computer as described in 11th invention as a client computer,
The character string contained in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and for each decomposed word, each part of speech and usage form are associated with each other as grammatical information and managed. Morphological analysis means to perform,
Character string unit recognition means for recognizing each character string unit of words, phrases, and sentences from the character string based on the grammatical information;
Keyword setting means for dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
Inquiry result returning means for inquiring the search keyword into a predetermined database and returning the inquiry result to the client computer.

A fifteenth aspect of the invention is installed in a computer having a display function of an acquired document and a communication function with a server computer that executes a morphological analysis process.
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting step for extracting a character string unit and setting a character string description portion of the extracted sentence as a keyword extraction range;
An analysis request message sending step for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
When a message including a search keyword is returned from the server computer that has received the analysis request message, a search result presenting step for inquiring the search keyword to a predetermined database and displaying the inquiry result;
This is an information retrieval support program that executes

A sixteenth aspect of the invention is installed in a computer having a display function of an acquired document and a communication function with a server computer that executes a morphological analysis process.
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting step for extracting a character string unit and setting a character string description portion of the extracted sentence as a keyword extraction range;
An analysis request message sending step for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
A search result presentation step of displaying and outputting a search keyword query result returned from the server computer that has received the analysis request message;
The information retrieval support program is executed.

  In the fifteenth or sixteenth invention, an information search support program for setting a character string range of a keyword extraction range according to the number of times a predetermined user input is continuously received within a predetermined time in the keyword extraction range setting step The seventeenth invention is provided.

An eighteenth aspect of the invention is installed in a server computer that can communicate with a client computer accessed via a network, and the server computer includes:
The character string included in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and the parts of speech and usage forms of each decomposed word are managed in association with grammatical information. A morphological analysis step;
Based on the grammatical information, from the character string, a character string unit recognition step for recognizing each character string unit of a word, a clause, and a sentence;
Keyword setting step of dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
A search keyword inquiry message sending step that includes the search keyword and returns a keyword inquiry message including an instruction to make the keyword inquired to a predetermined database to the client computer;
Is an information retrieval support program for executing

The nineteenth invention is installed in a server computer capable of communicating with a client computer accessed via a network, and the server computer includes:
The character string included in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and the parts of speech and usage forms of each decomposed word are managed in association with grammatical information. A morphological analysis step;
Based on the grammatical information, from the character string, a character string unit recognition step for recognizing each character string unit of a word, a clause, and a sentence;
Keyword setting step of dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
An information search support program for executing a query result returning step of querying the search keyword to a predetermined database and returning the query result to the client computer.

The twentieth invention is installed in a computer equipped with a browser, and
Accepts user input indicating the display position of a hyperlink to another web page set in the web page acquired by the browser, and executes a process of displaying and outputting the description content of the other web page as a summary sentence An information retrieval support program that
A linked document acquisition step of acquiring a document included in another Web page without displaying and outputting the document to a user;
An analysis target sentence specifying step for recognizing a character string unit based on a punctuation mark, a punctuation mark, or a line break as a sentence in the acquired document, and specifying a sentence including a predetermined number of characters or more as an analysis target sentence for the recognized sentence When,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, a morphological analysis step of managing the corresponding part of speech and usage as grammatical information,
Based on the grammatical information, extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words, and a word appearance number acquiring step of acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. An evaluation value acquisition step for acquiring
A summary sentence extraction step for extracting an analysis target sentence corresponding to the highest evaluation value as a summary sentence;
The information retrieval support program executes a summary sentence display step of displaying and outputting the summary sentence extracted in the summary sentence extraction step in a display area different from the Web page document acquired by the browser.

The twenty-first invention is installed in a server computer capable of communicating with a browser accessed via a network, and the server computer includes:
Acquires a Web page according to the URL sent from the browser, recognizes a character string unit based on a punctuation mark, a punctuation mark, or a line break in a document included in the Web page as a sentence. An analysis target sentence specifying step for specifying a sentence including the character as an analysis target sentence;
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, a morphological analysis step of managing the corresponding part of speech and usage as grammatical information,
Based on the grammatical information, extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words, and a word appearance number acquiring step of acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. An evaluation value acquisition step for acquiring
A summary sentence extraction step for extracting an analysis target sentence corresponding to the highest evaluation value as a summary sentence;
A summary sentence returning step of returning the summary sentence extracted by the summary sentence extracting step to the browser;
The information retrieval support program is executed.

According to a twenty-second aspect of the present invention, a browser is installed, a user input indicating a display position of a hyperlink to another Web page set in the Web page acquired by the browser is received, and the description of the other Web page is described. A computer that displays and outputs the contents as a summary sentence,
A linked document acquisition means for acquiring a document contained in another Web page without displaying and outputting it to a user;
In the acquired document, a character string unit based on a punctuation mark, a punctuation mark, or a line feed is recognized as a sentence, and a sentence including a predetermined number of characters or more is identified as an analysis target sentence for the recognized sentence. When,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, morphological analysis means that associates and manages each part of speech and usage as grammatical information,
Based on the grammatical information, the word appearance frequency acquisition means for extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words and acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. Evaluation value acquisition means for acquiring
Summary sentence extraction means for extracting the analysis target sentence corresponding to the highest evaluation value as a summary sentence;
Summary text display means for displaying and outputting the summary text extracted by the summary text extraction means in a display area different from the document of the Web page acquired by the browser;
It has.

A twenty-third invention is a server computer in a server-client system,
Means for communicating with a browser accessed over a network;
Acquires a Web page according to the URL sent from the browser, recognizes a character string unit based on a punctuation mark, a punctuation mark, or a line break in a document included in the Web page as a sentence. An analysis target sentence specifying means for specifying a sentence including a character as an analysis target sentence,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, morphological analysis means that associates and manages each part of speech and usage as grammatical information,
Based on the grammatical information, the word appearance frequency acquisition means for extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words and acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. Evaluation value acquisition means for acquiring
Summary sentence extraction means for extracting the analysis target sentence corresponding to the highest evaluation value as a summary sentence;
Summary sentence return means for returning the summary sentence extracted by the summary sentence extraction means to the browser;
It has.

  The present invention extends to a program storage medium storing the information search support program described in any of the above inventions, and this program storage medium is the twenty-fourth invention.

  According to the information search support program and the computer having the information search support function of the present invention, an appropriate word is extracted as a search keyword from a document described in a non-separated language, and information search is performed based on the word. The search result can be presented to the user. In addition, before instructing a hyperlink set in a web page and obtaining another web page, the contents of the other web page can be confirmed. Thereby, unnecessary hyperlink instruction operation can be eliminated, and time and labor required for information retrieval can be saved.

It is the schematic which shows the result of having performed the morphological analysis of a certain document by the information search assistance program in the Example of this invention. It is the schematic of the process which extracts a clause from the sentence by the said program. It is the schematic of the process which recognizes the dependency of the clauses by the said program. It is the operation screen of the said program displayed on a computer. It is the schematic of the link reset process of the web page by the said program. It is the schematic of the verb final form conversion process by the said program. It is the schematic of the adjective end form conversion process by the said program. It is the schematic of the end form conversion process of an adjective verb by the said program. It is the schematic of the word appearance frequency presentation function by the said program. It is a flowchart of the process by the information search assistance program which operate | moves on a server client system. It is the schematic of the screen which the client computer which comprises the said server client system displays. It is the schematic of the screen which the client computer which comprises the said server client system displays. It is the schematic of the screen which the client computer which comprises the said server client system displays. It is the schematic of the screen which the client computer which comprises the said server client system displays. It is the schematic of the screen which the client computer which comprises the said server client system displays. It is the schematic of the screen which the computer in which the said program was installed displays. It is the schematic of the screen which the computer in which the said program was installed displays.

=== Embodiment of the Invention ===
As an embodiment of the present invention, a personal computer (PC) on which an information search support program (hereinafter referred to as a keyword extraction program) according to the present invention is implemented is exemplified. The keyword extraction program of the present invention extracts an appropriate character string as a search keyword from a document given as a processing target, inquires the search keyword to a search engine or a dictionary / dictionary site, and presents the inquiry result. Execute the process. The document to be processed may be simple text data that does not include a format as long as it can be browsed by a computer, or may be a data format for various word processing software. The keyword extraction program in this embodiment functions as a unique Web browser or a plug-in of an existing browser, and analyzes a document described in a Web page obtained by the browser. That is, a document included in data described in HTML is a processing target.

=== Morphological analysis / Syntactic analysis ===
The keyword extraction program decomposes the document into words by a well-known morphological analysis. Further, the document is parsed to recognize the dependency relationship in the document. A search keyword can be specified in units of words, phrases, and sentences by user input.

  In morphological analysis, a dictionary in which part-of-speech and utilization forms are associated with a large number of words in an external memory of a PC is prepared, and a keyword extraction program refers to the dictionary to extract words from a document to be analyzed. Go. Then, each extracted word is managed in association with grammatical information such as part of speech and usage form. FIG. 1 shows a schematic diagram of the result of morphological analysis of a certain sentence. In this example, documents such as “Mr. Noda, who stayed at the International Space Station conducted a third outboard activity early in the morning of the 3rd” were broken down into words, and grammatical information was associated with each word. Yes.

  In this embodiment, each character string unit of a word, a phrase, and a sentence is recognized from a character string continuous in the document, and the dependency relation is recognized by parsing the phrase in the recognized sentence. A phrase is a phrase that is a string unit that makes sense by adding “ne” to the end of the string. Specifically, the sentence “I read a book” is a word obtained by morphological analysis. If the word “ne” is added to the end of consecutive words before and after it, it makes sense. It becomes. That is, since each character string makes sense such as “I am I”, “I am reading a book” and “I read it”, each of “I am”, “I read a book” and “Read” is a phrase. “I read a book” and “read a book” are also phrases. In this embodiment, phrase breaks are recognized based on combinations of parts of speech of successive words in the sentence. The dictionary referred to by the keyword extraction program includes a table that associates the combination type of parts of speech before and after the sentence with the determination of whether or not the phrase is a phrase. Note that the “sentence” is a character string delimited by punctuation marks and punctuation marks in this embodiment. Furthermore, the keyword extraction program recognizes the dependency relationship in the sentence by syntactic analysis. Dependent relationships include, for example, a combination modification relationship such as “delicious meal”, a combined modification relationship such as “run slowly”, and a subject predicate relationship such as “I walk”.

  FIG. 2 shows a recognition procedure for each character string in this embodiment. Here, the phrase recognition procedure in sentence 10 “I speak fluent Japanese” is shown. First, the sentence 10 to be analyzed is subjected to morphological analysis, and the part of speech 12 of each word 11 is acquired (A). Next, referring to the table 16 in which the type 14 corresponding to the combination (13a, 13b) of the part of speech 12 (13a, 13b) of the word 11 preceding and following in the sentence 10 is associated (B), Which type 14 in the table 16 corresponds to each boundary 17 between words is specified (C). Then, it is determined from the determination result 15 corresponding to the identified type 14 whether adjacent words form one phrase or not, and the minimum unit phrase 18 is cut out from the sentence (D).

  Further, the keyword extraction program recognizes the dependency relationship between clauses in the sentence 10 divided for each clause 18 of the minimum unit by the syntax analysis. FIG. 3 shows a dependency recognition procedure by syntax analysis. In the sentence 10 divided into the minimum unit clauses 18 (A), which clause 18 is related to which clause 18 and what dependency relationship (19a-c) is obtained (B), the adjacent clause 18 is related. If it is in the receiving relationship (19a-c), it recognizes that it can become a bigger phrase (18b, 18c) by the adjacent phrase 18 (C) (D). Thus, in this embodiment, when a sentence is divided into a plurality of character string units, the character position to be divided, such as for each word, for each minimum unit clause, and for each of a plurality of clauses having a dependency relationship, etc. Can be set freely.

=== Outline of Keyword Extraction Program ===
The keyword extraction program of this embodiment analyzes the HTML of a Web page obtained by the browser and extracts a document part to be processed. Then, the above-described morphological analysis and syntax analysis are applied to the document, and character strings corresponding to words, phrases, and sentences in the document recognized based on the analysis result are extracted as search keywords. Then, the search keyword is referred to a search keyword inquiry destination (search engine, dictionary / encyclopedia site, etc.) set by user input.

  Specifically, for a document included in a Web page received by the browser, the hyperlink (link) that was intentionally set by the Web page creator in the original document is discarded while maintaining the display state. Then, an HTML is newly created by setting a link indicating that the character string is provided to a predetermined inquiry destination at an appropriate word in the document or a character string description location such as a clause or sentence. In this embodiment, the keyword extraction program manages a URL description format for giving a search keyword for a plurality of reference destinations and character code designation information when the reference destination recognizes Japanese. In the character string portion for setting a link, the character string is converted into a specified character code and described according to a specified URL description format.

  For example, a URL for giving a search keyword to a certain search engine is “http://search.engine.co.jp/bin/query?=” followed by a search keyword, and the search engine is EUC. When accepting Japanese (Kana-Kanji) by code and setting a link to the word entry “book” in the document, the URL to give the search keyword to the search engine is “http://search.engine.co.jp / bin / query? =% CB% DC ", and the link to be set including the HTML tag is" <a href = "http://search.engine.co.jp/bin/query?=% CB% DC "> Book </a>".

  When a character string corresponding to Japanese (Kana-Kanji) is described in the URL, it is necessary to convert the character string into a well-known URL encoding format. That is, a Kana-Kanji character described in 2 bytes is decomposed into 1-byte characters and displayed in hexadecimal, and the Kana-Kanji character is converted by describing the character in hexadecimal notation after the “%” symbol. In the previous URL “http://search.engine.co.jp/bin/query?=%CB%DC”, “% CB% DC” is a URL-encoded version of “book”.

=== Extraction of search keywords ===
The keyword extraction program of this embodiment uses a suitable character string in a document described on a Web page as a search keyword, and supports user information search using the keyword as a starting point. And the function for extracting a search keyword appropriately is provided so that the user can acquire the target information with high probability. For example, the link setting part is an entire sentence delimited by punctuation marks, and the entire sentence can be clicked with a mouse, that is, while the entire sentence is an instruction target for inquiring to a predetermined search engine or a predetermined dictionary site, If you place the mouse cursor on the sentence and click it, not only the sentence itself is set as a search keyword, but if the query target of the search keyword is a search engine, multiple phrases or multiple phrases included in the sentence A word can be set as a search keyword for AND search. That is, each word, phrase, and sentence in the document can be freely set as a mouse click target. Even if the object of the mouse click is a sentence, what is actually queried to the search engine as a search keyword is the sentence itself, or a plurality of words included in the sentence used for AND search Or multiple phrases. Hereinafter, the function of the keyword extraction program of the present embodiment will be described in detail by giving an example of inquiring a search engine for a search keyword extracted from a document.

=== User interface ===
FIG. 4 shows an operation screen of the keyword extraction program that is running on the PC. In this example, the keyword extraction program functions as a plug-in of Internet Explorer (registered trademark) serving as a browser, and an area 31 for operating the keyword extraction program is displayed in a window frame 30 of the browser. . A plurality of screens 32 can be selected and displayed by tabs 33. The plurality of screens 32 can be used to display a web page obtained by the browser as it is, and a document included in the web page in various ways. And each screen corresponding to each analysis result is included. Of course, the user interface environment of the program can be set as appropriate, such as displaying a plurality of screens 32 side by side. In this figure, the screen of the document in which the link to the search engine is reset is selected and displayed.

  In the program operation area 31, a menu 34 and a toolbar 35 for operating the program are displayed. For example, the tool bar 35 includes a list down box 36 for selecting a unit of a character string to be clicked (word, phrase, sentence), and a list down box for selecting various settings related to search keyword configuration. 37 etc. are arranged. The user operates the keyword extraction program to extract a search keyword that matches his / her information search purpose from the document described on the original Web page, and browses the result of the actual information search based on the search keyword can do. Of course, if the information obtained by the information search is a document, the link can be reset for the information search for the document.

=== Link setting by word ===
The most basic unit to be clicked is a word. 5A and 5B illustrate a process of setting a word included in a Web page document as a click target and a search keyword. The document 40a on the original Web page is underlined at the position where the character string to which the link is set. This link is intentionally set by the Web page creator, and usually a specific Web page is the link destination. In this example, a link is set in the description section of “International Space Station” in Document 40a, “Mr. Noda, who is staying at the International Space Station, conducted a third outboard activity early in the morning of 3rd.” (A).

  The keyword extraction program applies morphological analysis to the document in the Web page, sets a link to the search engine at the character string description location of the word included in the document 40a, and underlines the reset location of the link. A new HTML is generated by pulling, and the HTML is displayed by the browser. As a result, the original link is destroyed, and “International”, “Space”, “Station”, “Nii”, “Stay”, “Medium”, “No”, “Noda”, “Samurai”, “San”, “Ra”, “Ga”, etc. Thus, the document 40b in which the link is reset for each word is displayed in the form of the Web page in the operation area of the program (B).

  In addition, for a character string (compound noun) using consecutive nouns, the entire compound noun can be set as a click target. For example, in the sentence example shown in FIG. 5, rather than giving each of “International”, “Space”, and “Station” as individual search keywords to the search engine, the user is more interested in using “International Space Station” as a search keyword. In some cases, it is easy to obtain information. In this way, the user can set a character string (compound noun) of consecutive nouns as one click target and a search keyword according to his / her search purpose. In setting this compound noun, for consecutive nouns, the number of nouns to be included in the compound noun from the beginning ("International universe" and "Station" where the number of consecutive nouns is 2), or consecutive nouns It may be possible to designate the number of nouns from the end of the name ("international" and "space station", where the number of consecutive nouns is two). Of course, if a dictionary used for morphological analysis includes both a noun and a compound noun that includes that noun, such as “Universe” and “Space Station”, the compound noun is preferentially adopted. There may be compound nouns that are extracted automatically.

  As for the timing to give a word to the search engine, when the user clicks a certain word, the word may be given to the search engine immediately, or the word is clicked and the word is set as a search keyword. Then, a button or the like that accepts a search instruction by clicking again may be arranged in the operation screen generated by the keyword extraction program. Then, when a plurality of words are selected one after another and the search instruction is accepted, the keyword extraction program executes an AND search using the selected plurality of words. As is well known, an AND search using a plurality of search keywords may be performed by enumerating a plurality of search keywords via a half-width space or a full-width space and providing the result to the search engine.

=== Link setting by phrase ===
The keyword extraction program can set a phrase obtained from a word string obtained from the result of morphological analysis as a search keyword and set it as a click target. For example, in the case of the sentence “I went to school in a hurry”, the phrases “I”, “To school”, “By bicycle”, “Hurry” and “I went” would become clickable. If you click on a phrase, the string of that phrase becomes the search keyword.

  Furthermore, it is also possible to click on a clause that has an adjacent linkage modification relationship. In the case of the sentence “I eat delicious meals,” considering that there are many requests for information retrieval in units such as “delicious meals,” the clauses that are adjacent to each other are modified together. It is possible to click by phrase. In other words, it can be understood that the phrase “delicious” is related to the phrase “meal” by a syntactic modification, so that the character string corresponding to the phrase “delicious meal” may be clicked. If a phrase is set as a unit of the search keyword, in this example, three character strings “I am”, “Delicious meal”, and “Eat” are clicked respectively.

Of course, it is also possible to click on a phrase that is in an adjacent continuous modification relationship. In the case of the sentence “I walk slowly,” a character string such as “walk slowly” may be appropriate as a search keyword. For this reason, it is possible to click adjacent clauses that are in a continuous modification relationship as a group. In this example, two character strings “I am walking slowly” are clicked as search keyword candidates.

=== Link setting by sentence ===
In some cases, you may want to search for information using punctuation marks in Japanese kanji or mixed text, or a character string between them. For example, suppose there is a sentence “I want a house where all the family can enjoy in a quiet atmosphere and can relax slowly.” If this sentence is simply divided into units of punctuation marks, it can be further divided into two character strings: “A family that everyone can enjoy in a quiet atmosphere” and “I want a home where people can relax slowly.” The keyword extraction program can appropriately set a search keyword based on the results of morphological analysis and syntax analysis even when searching for information using such a long character string.

  In the information search in which the previous word or clause is clicked, information including a character string is mainly searched for directly. On the other hand, in the information search in which the sentence is clicked, it is assumed that information related to the meaning and content included in the character string is searched. In this embodiment, a document is set as a click target by a user input, and various search keyword forms to be extracted from the click target sentence are set by a user input. it can. Thereby, the user can obtain information related to the content of the sentence with high accuracy.

<Entire sentence>
The simplest form of use of a search keyword extracted from a sentence is a form of “passing a sentence character string directly to a search engine”. For example, if the sentence is "I want a house where all the family can enjoy in a quiet atmosphere and want a relaxing home," this string itself becomes the search keyword. Of course, when a long sentence is used as a search keyword as it is, the possibility that related information is shown as a search result is low. However, it is effective when searching directly for an article that is the source of the character string.

<Words and phrases>
The next simple usage form of a search keyword extracted from a sentence is a form of “AND search is performed using a word extracted by morphological analysis in the character string of the sentence”. If this form is specified by user input, the keyword extraction program will use the previous sentence “I want a house where everyone in the family can enjoy in a quiet atmosphere and relax.” "Everyone""Ga""Quiet""Atmosphere""In""Enjoy""House""De""Yes""Slow""Relax""Like""Housing""Ga""Iwant" Set a word as a search keyword for AND search. It is also possible to set a search keyword by phrase, in which case the same sentence is "family""everyone""quiet""atmosphere""enjoyable""home""slow" When the sentence is divided into the phrases “Relaxing”, “Housing is”, “Want”, and clicking on the sentence, an AND search is executed using the character string of each phrase included in the sentence as a search keyword.

<Part of speech specification>
When searching for information related to a very long character string, it may be easier to obtain desired information if a search keyword is set by extracting or excluding only words of a predetermined part of speech from the sentence. In this embodiment, the grammatical information is associated with each word obtained by morphological analysis, so that the search keyword can be set by excluding only the part of speech specified by the user. For example, when excluding particles and auxiliary verbs, the sentence “Family,” “Everyone,” “Quiet,” “Atmosphere,” from the sentence “I want a house where everyone in the family can enjoy a quiet atmosphere and relax.” An AND search is performed using “enjoyable”, “home”, “present”, “slow”, “relax”, “housing”, and “want” as search keywords. In addition to particles and auxiliary verbs, excluding verbs (verbs, adjectives, adjective verbs), that is, selecting only important parts of speech such as nouns, "family""everyone""atmosphere""house""slowly"" An AND search is performed using “house” as a search keyword. Note that when setting a search keyword as a click target of a sentence, a noun may be selected and set as a compound noun or a normal noun by user input === Terminal conversion ===
For example, in a dictionary / encyclopedia site, search terms for prescriptive terms are basically accepted in a closed form. For this reason, when trying to query a dictionary / encyclopedia site for words in a document, if the query is made as it is in the document, the corresponding index cannot be found and the information that is the query result cannot be obtained. . Conventionally, when a user tries to look up the vocabulary of a term that is not a terminal form on a dictionary / encyclopedia site, the user converts the phrase's notation in the document into a terminal form, and after the conversion It was necessary to query the character string of as a search keyword. In addition, there is a problem that in terms of phrases including kanji, the user cannot convert to the final form without knowing how to read the kanji. Therefore, when the character string to be used as a search keyword is a predicate, the keyword extraction program of this embodiment corrects the character string to an end form and sets the character string as a search keyword for actually inquiring. It has an end-to-end conversion function.

  In this embodiment, each word decomposed by morphological analysis is managed by associating grammatical information such as parts of speech and inflection forms. Therefore, the keyword extraction program can also acquire a usage form other than the notation if a prescription written in a certain usage form exists in the document. Fig. 6 shows how to use the "La line five-step verb" called "Run". For example, if the verb “run” is described in the document as “not run”, the keyword extraction program breaks it down into three words (morphemes) “run” “ra” “no” It recognizes that the grammatical classification of each morpheme is “La line five-stage verb stem”, “La line five-stage verb inflection ending form”, and “auxiliary verb”. And by using the morpheme of the stem `` Run '' (La line five-step verb stem) and the final form of `` Lu '' five-stage verb ending, `` Run '' and `` Ru '' (the final form of the La line five-step verb) The final form of “Run” is generated.

  The same applies to adjectives. FIG. 7 illustrates a utilization form of “beautiful”. Recognize the stem "Beauty" by morphological analysis in any usage form, and generate the final "Beautiful" by combining this stem and the adjective end use "I" Can do. As for the adjective verbs, the use form is shown in FIG. In the same way as verbs and adjectives, the stem “ambiguous” and the adjective verb ending “da” can be combined to generate the final form “ambiguity”. In this embodiment, it is also possible to select a part of speech to be converted to an end form by user input. This is because when a predicate is given to the search engine, the target information may be obtained by inquiring the character string as written, rather than the end form.

=== Related Document Search Function ===
As described above, the keyword extraction program of the present embodiment is designed to obtain information intended by a user who browses a document by skillfully applying natural language analysis techniques such as morphological analysis and syntax analysis to the WWW mechanism. The search keywords can be extracted with high accuracy. Further, in this embodiment, as a result of the natural language analysis of the document, a related document search function for acquiring the appearance frequency of the word in the document and searching for other information related to the content of the document based on the appearance frequency is provided. ing.

  For example, “Satoshi Noda and others who are staying at the International Space Station conducted a third outboard activity early in the morning on the 3rd. In this activity, Robinson was the first ever space shuttle in space. Noda helped with the repair, and the Space Shuttle re-entered the atmosphere, causing the temperature to rise locally in the vicinity of the overhang and jeopardize safe return, so the US Aerospace Agency will carry out the repair. Prior to the repair of the space shuttle, Mr. Noda also carried out work to install space experiment equipment and an outboard storage in the ISS, with a total of 7 hours for extravehicular activities. When trying to obtain information related to the document "External activities were also carried out on the 30th and August 1st", the grammatical information and the appearance frequency are associated with the words in the document, and the appearance frequency is In descending order It is giving to the predetermined search engine as a search keyword a word. FIG. 9 shows a list of words in the document of the above example. Each word is associated with grammatical information, the number of appearances, and the appearance frequency. In the present embodiment, nouns, sub-variables, and proper nouns are extracted as words to be counted for the number of appearances, and the top six words having the highest appearance frequency are set as search keywords. In addition, the search keyword that is actually inquired to the search engine is a character string that is converted to a final form for the verbs (verb, adjective, adjective verb). Of course, the part of speech to be extracted, the necessity of the final form conversion, the number of search keywords to be inquired, and the like can be set by user input.

=== About incidental functions ===
The keyword extraction program allows users to select appropriate search keywords, understands the content of documents, in addition to the function of querying search engines etc. for search keywords extracted based on natural language analysis technology, or It is equipped with various incidental functions for performing information retrieval work with a light user interface environment. In addition, the result of natural language analysis applied to a document and the result of HTML analysis as preprocessing for the analysis are utilized to make the structure of HTML in a Web page and the language structure in a document in the Web page. Etc. It also has an incidental function to present to the user. Hereinafter, an incidental function in the keyword extraction program of this embodiment will be described.

=== Summary function ===
A sentence summarizing the contents of a document seems to contain more words (important words) that appear frequently in the document. Therefore, based on the appearance frequency of words in the document to be processed, sentences containing many important words with high appearance frequencies are extracted and presented as summaries. In this embodiment, there are two types of cases where, by user input, sentences containing many important words are presented in the order in the original document, and sentences are presented in order from sentences having a high proportion of important words. It can be presented by the method.

=== Appearance frequency presentation function ===
A list of appearance frequencies of words created by the related document search function can be displayed as it is. The posting order of words in the list can be set in the order of appearance frequency, in the order of the longest notation of characters, and in the order of character codes by user input. Similarly to the previous related document search function, the part of speech of the word whose appearance frequency is to be counted can be set by user input. In this embodiment, words that are not registered in the dictionary that the keyword extraction program refers to in the morphological analysis can be separately counted as new words. In addition to new words, words with only katakana notation, words with katakana notation, words with only alphabetic notation, words with alphabetic notation, words containing numbers or numbers (or words suitable for recognition as numbers), pronouns (he , I, this, it ...), nouns such as numbers related nouns (1st, 1st year ...), nouns related to time series (today, tomorrow ...) can be excluded or counted. In particular, numeric related nouns may suggest specific events and are likely to be words that characterize the document. Of course, it may not make sense at all. Katakana and English words include coined words and foreign words, so users can know unknown words for themselves by counting and presenting them individually.

  Note that the appearance frequency of verbs, adjectives, and adjective verbs can be determined by user input while maintaining the inflected form in the sentence or after unifying it into a final form. For compound nouns, the number of consecutive nouns constituting the compound noun can be specified by user input.

=== HTML analysis function ===
An HTML analysis function for analyzing the HTML describing the Web page obtained by the browser from various perspectives and presenting the analysis result is provided. In this embodiment, a document extraction function, a tag display function, and a link presentation function are provided.

  The document extraction function is a function for extracting and displaying only a “text” portion from which tags and image-related information in HTML are removed. For example, an advertisement or the like placed in a Web page prevents a user from recognizing which document is important and which document is unnecessary in the page. In the present embodiment, not only HTML tags are excluded, but the number of consecutive characters between the tags can be counted, and only characters greater than the number specified by user input can be displayed.

  The tag display function presents a list of HTML tags by type. The link presentation function presents a list of link destination URLs set in HTML.

=== Language structure presentation function ===
The keyword extraction program presents the language structure of the document to the user. The ruby function that assigns ruby to the kanji in the document, the kana conversion function that converts all words to Hiragana, and the words in the document by part of speech. A part-of-speech identification display function that displays in a way that can be identified, a segmentation that uses words obtained from morphological analysis results, and a display that shows which clauses and what clauses have a relationship with each clause It has a relationship display function.

  The ruby-functional conversion function enables even children who cannot read kanji to read documents. Moreover, if a document with only kana characters obtained by the kana conversion function is given to known voice conversion software, even a visually impaired person can read the document by “listening”.

  According to the part of speech identification display function, the user can intuitively understand the language structure of Japanese. That is, the appearance frequency of parts of speech, the arrangement in a sentence, the context of different parts of speech, the continuous / discontinuous relationship, and the like can be sensed. The part-of-speech identification display function enables visual distinction of parts of speech by changing the color of each word decomposed by morphological analysis for each part of speech. If used in the field, it is effective for grammar education and can be expected to improve writing skills.

  The dependency display function is based on the grammatical information of each word obtained by the above morphological analysis and syntactic analysis, the clause recognized from the grammatical information, the context of each clause, etc. And which clauses are in what relationship. As described above, types of dependency relationships include combination modification, continuous modification, subject predicate, and the like. In this embodiment, the dependency relationship between phrases is displayed in different colors depending on the type. By identifying and displaying the dependency analysis results between clauses in this way, the relationship between clauses can be intuitively recognized, and the dependency display function in the keyword extraction program function should also be used in the field of Japanese language education. Can do.

=== Processing timing of morphological analysis and character string to be processed ===
In the above embodiment, morphological analysis is performed on all documents in the page when the browser obtains the Web page. For example, the keyword extraction program may perform morphological analysis when a word in a Web page is selected. By doing so, it is possible to minimize the occurrence of the morphological analysis process and avoid applying an extra load to the PC when a Web page is obtained for the purpose of browsing only.

  In addition, if the character string that is meaningful as a sentence is processed, such as the character string delimited by punctuation or punctuation from the character string before and after the selected word, the character string to be analyzed is not processed as a whole document. The amount of information to be reduced is reduced, and the processing result by the morphological analysis can be presented in a shorter time. For example, when a character string unit or part of speech to be a search keyword is specified, if an arbitrary character position in the document is specified, that character is included based on the surrounding punctuation of the character at the specified position. The sentence is targeted for analysis, and the sentence is subjected to morphological analysis and syntax analysis. Then, the characters before and after the character at the designated position are extracted by the designated character string unit, and the extracted character string is used as a search keyword.

  For example, the previous document “Mr. Noda, who is staying at the International Space Station, conducted the third outboard activity early in the morning on the 3rd. The space shuttle was repaired and supported by Mr. Noda because the space shuttle would re-enter the atmosphere and the temperature would rise locally in the vicinity of the overhang, causing a safe return. The space agency decided to carry out the repairs, and prior to the space shuttle repairs, Mr. Noda also carried out work to install space experiment equipment and outboard storage in the ISS. As for the second sentence "Robinson was the first space ever in space for this day's activities," the two conducted extravehicular activities on July 30 and August 1. Repair the shuttle Noda's is that it has support it. "" Scan "has been instructed to. In this case, a sentence delimited by a punctuation mark at the stage where "su" was indicated, "In this day's activities, Robinson repaired the first space shuttle in outer space, and Noda supported it. ”Is analyzed. Here, if the character string unit to be used as a search keyword is a word, “space shuttle” is given to the search engine. If the character string unit is a sentence and a noun is specified as the part of speech of the search keyword, the nouns “first in history”, “space”, “space shuttle”, and “repair” included in this sentence are given to the search engine.

=== How to specify search keyword extraction range ===
An example of an operation method for designating a range from which a search keyword is extracted (a keyword extraction range) in a document will be described. As a keyword extraction range, a method of first specifying a range of an arbitrary character string by a known drag operation and specifying a word or compound word within the range as a search keyword is conceivable. In addition, the keyword extraction range is sequentially expanded according to the number of consecutive predetermined operations, such as compound words, sentences, and documents, depending on the number of times a predetermined operation is repeated with an arbitrary character description position in the document indicated. There are ways to go.

  As this predetermined operation, a mouse click operation can be considered. Taking the previous document “Staying at the International Space Station ...” as an example, for example, if a single click operation is performed at the character position of “Navigation”, the character “Aviation” is determined from the characters before and after the character position. A word is extracted, and only this word “aviation” becomes a keyword extraction range. That is, the word “aviation” itself is a search keyword.

  If you double-click at the same position, the keyword extraction range will be “Aerospace Organization” as a compound word and “US Aerospace Organization” as a triple click. Of course, if only “US Aerospace Organization” is registered in the dictionary, “US Aerospace Organization” becomes the keyword extraction range by double-clicking. The search keyword when the compound word is the keyword extraction range is, for example, “Aerospace Agency” is a keyword extraction range, each word “aviation” “space” “mechanism” becomes a search keyword, The term “Aerospace Organization” will be used as a search keyword.

  Furthermore, if the click operation is continued, the sentence containing the letters “Kai” “If the space shuttle re-enters the atmosphere, the temperature rises locally near the overhanging area, and safe return is jeopardized. “Aerospace Agency has decided to carry out repair” is the keyword extraction range, and the search keyword is specified from this sentence according to the prior settings. Of course, before the sentence is set as the keyword extraction range, the character string “American Aerospace Organization has decided to carry out repair” delimited by punctuation marks may be set as the keyword extraction range.

  The search keyword is not limited to only a noun, and may be specified by a phrase including a verb or an adjective in the noun, or an adjective or a verb end form may be specified as a search keyword. If the number of clicks is further increased, the entire document becomes the keyword extraction range. The keyword extraction range may be cycled according to the number of clicks, such as word → compound word → sentence → document → word → compound word. In addition, as the order of expansion of the keyword extraction range, a clause can be set between a compound word and a sentence, or a paragraph can be set between a sentence and a document. For example, depending on the number of clicks, a word The keyword extraction range may be expanded in the order of compound word → clause → sentence → paragraph → whole document. As a predetermined operation when setting the keyword extraction range, an appropriate operation such as pressing a specific key in addition to a mouse click can be considered.

  In this way, by setting the keyword extraction range according to a predetermined continuous operation, it is not necessary for the user to move the cursor or the like from the description position of the character that the user wants to specify as the search keyword. Therefore, the search keyword intended by the user is surely given to the search engine, and by expanding the keyword extraction range, the search keyword is selected from the character strings around the intended search keyword, The possibility that a record related to the description content of the keyword extraction range is acquired can be increased.

  As specific processing for setting the search keyword extraction range and search keyword according to successive click operations, morphological analysis is performed when a document or sentence is analyzed in advance and when the click operation is performed. A case may be considered. When a document or sentence is analyzed in advance, the keyword extraction program manages the correspondence between the number of clicks and the keyword extraction range, and composes words in the characters before and after the character description position where the click operation is accepted. A character string may be acquired, and a compound word or sentence including the word may be set as a search keyword extraction range according to the number of clicks.

  On the other hand, when morphological analysis is performed when a click operation is performed, the keyword extraction program manages the correspondence between the number of clicks and the keyword extraction range, and characters before and after the character description position where the click operation is accepted. Punctuation marks and punctuation marks are acquired from the sequence, and the punctuation marks or character strings delimited by the punctuation marks are extracted as morphological analysis targets.

  Then, a keyword extraction range corresponding to the number of consecutive click operations is specified from the character string before and after the character description position where the click operation is accepted, and a search keyword is specified from the extracted range according to the user's setting. When a click operation is performed a predetermined number of times or more, the entire document may be a morphological analysis target. It is also possible to present to the user which character string portion of the document being displayed the search keyword is extracted, such as by highlighting the character string in the keyword extraction range.

=== Server / Client System ===
In the above embodiment, the keyword extraction program installed in the PC includes all functions such as the morphological analysis function and the function of giving the search keyword to the search engine. The keyword extraction program is not limited to this example, and may be a program that operates in cooperation between a client computer and a server computer in a server / client system. That is, the PC is a client computer, and the client computer has a function as a user interface of the keyword extraction program. On the other hand, the server computer that communicates with the client computer is provided with a morphological analysis processing function and a dictionary used for the processing. Then, the server computer performs morphological analysis on the document being displayed on the client computer.

  Hereinafter, an embodiment of the information search support function by this server / client system will be exemplified. In the example shown here, the client computer is a PC on which a browser is installed, and the keyword extraction program installed on the PC side is installed as a browser plug-in. Here, the processing function of the keyword extraction program on the client computer side is called a client program, and the processing function on the server side is called a server program.

  The client program has a function of setting the search keyword extraction range and the search keyword part-of-speech from the document displayed by the browser and providing the setting information to the server program. Note that, as described above, the client program adopts a user interface that sets a keyword extraction range according to the number of consecutive mouse clicks. The keyword extraction range may be expanded to a document, but in this example, in order to reduce the communication load between the client computer and the server computer, a sentence consisting of character strings separated by punctuation marks is used. Is the largest morphological analysis target character string. The keyword extraction range is expanded in the order of word → compound word → sentence.

  The client computer stores location information on the network of the server computer that is a communication partner, such as a URL, as client program installation information, and the client program is executed as a plug-in of a running browser. When there is a predetermined user input, a part of the document being displayed on the browser is sent to the server computer for morphological analysis, and various processes based on the analysis result are executed.

  FIG. 10 illustrates a flow of processing when the client program and the server program operate cooperatively. In addition, FIGS. 11 to 15 show transitions of screens displayed on the client computer along with the processing flow. In this example, a screen 80 for operating a client program incorporated as a browser plug-in is slightly different from the screen 30 shown in FIG. 4, and a radio button 81 for designating a search keyword reference destination. Or a radio button 82 that is selected when the ruby is shaken, a “related document search” button 83 for executing the related document search function, and a “word appearance frequency analysis” for analyzing the appearance frequency of each word included in the document. ”Button 84 and an“ option ”button 85 for setting various conditions such as specifying the part of speech of the search keyword or setting a search site or a dictionary site to which the search keyword is introduced. It is arranged in the display area of the operation screen 80.

  In the client computer 50, when a predetermined number of click operations are received on a displayed document in a state where, for example, a noun is set as a part of speech of a search keyword by a predetermined user input (FIG. 11), A click operation specified by the text data corresponding to the character string 87 between the preceding and following character points, the number of characters from the beginning of the character string 87, and the like are detected. A message (morpheme analysis request message) including information on the position where the request is received, the number of clicks, the part-of-speech specification information of the search keyword set in advance, and information on the query destination of the keyword specified by the radio button 81 ) Is sent to the server computer 60 (s1 to s7). Here, it is assumed that a Web page including a search keyword is searched, and a predetermined search engine 70 is set as an introduction destination of the search keyword.

  The server computer 60 performs morphological analysis on the sent text data, breaks it down into words, associates grammatical information with each word, and specifies a keyword extraction range based on the number of clicks from the client computer 50. The text data is formatted so that the character string in the keyword extraction range is highlighted. Further, the character string of the search keyword is extracted from the extraction range based on the character position information and the part of speech designation information. Then, the browser of the client computer 50 is redirected to a predetermined search engine 70 to generate a URL for giving a search keyword to the search engine 70, and the URL and the formatted text data are used for the client computer. 50 is returned (s8 to s13).

  The client computer 50 uses the client program to replace the character string that has been subjected to morphological analysis in the document previously displayed by the browser with the text data that has been formatted and returned by the server computer 60. The character string with the search keyword extraction range 88 highlighted is displayed (FIG. 12). Further, the URL describing the redirect to the search engine 70 is interpreted and a search keyword is given to the search engine 70. Then, the search result Web page returned from the search engine is displayed (FIG. 13, s14 to s18). In this example, a Web page describing the search result sent from the search engine 70 is displayed in a separate window 90.

  FIG. 12 shows an example in which a character string “space shuttle” is set in the keyword extraction range 88 by a double click operation. In the single click operation, “space” is the keyword extraction range (FIG. 14). In the click operation of 3 times, the sentence that is delimited by the punctuation mark is set as the keyword extraction range, which is the first time in the history of repairing the space shuttle in outer space. Will be displayed (FIG. 15).

  The server computer 60 may send only the search keyword to be given to the search engine 70 to the client computer 50 instead of the URL describing the redirect. Then, the search keyword received from the server computer 60 by the client computer 50 is given to the set search engine 70. Alternatively, the server computer 60 may be configured to transfer the search result returned by giving the search keyword to the search engine 70 to the client computer 50.

  In this way, it is possible for the server computer side to execute a process with a large load, for example, even in a keyword search using a mobile phone (browser phone) equipped with a browser having a processing capability inferior to that of a PC. It is possible to provide a light user interface environment to the user. In other words, by setting the browser phone as a client computer, it is possible to cause the server computer to execute a morphological analysis process with a large load on the document displayed on the browser phone. It is only necessary to execute a process with a light load that gives the search engine a keyword extracted based on the analysis result. If the server computer gives a search keyword to the search engine and transfers the search result to the browser phone, the processing load on the browser phone is further reduced.

=== Business model ===
As is well known, in a search site, a record corresponding to a search keyword received from a browser and an advertisement content (link to a website intended by the advertiser) associated with the keyword are set by a database attached to the search engine. Displayed on the search result Web page.

  In general, advertisement content is described on a search result Web page in preference to a record. Therefore, if there is any user input to the document displayed on the client computer, the advertising content is displayed on the client computer and is used for browsing by the user. For this reason, the client computer is instructed to set a link set in the advertisement content, and the user browses the website intended by the advertiser, or the user performs some transaction (response to questionnaire, product, etc.) on the website. Can be expected to generate). Therefore, server computer management operators can be expected to build a business model that obtains advertising revenue from search site management operators and advertisers for advertisement content displayed on client computers. .

=== Summary sentence presentation function of linked page ===
The summary function described above presents a summary sentence of the Web page that is currently displayed by the browser. According to the information search support program of the present embodiment, there is further provided a linked summary sentence presentation function for presenting a summary sentence for another Web page that can be acquired based on a hyperlink set in the displayed Web page. Can be realized. In the link destination summary sentence presenting function, when a hyperlink to another Web page (link destination page) is set on the Web page (link source page) that the browser is currently browsing, the link destination Instead of displaying the full text of the page, only the summary text of the linked page is presented to the user in advance.

  FIG. 16 shows an outline of the screen displayed on the display of the PC by the link destination summary sentence presentation function. In this embodiment, when a “mouse over” operation is performed in which the mouse cursor 93 is placed on the link location 92 set in the link source page 91 being displayed on the PC, the display area of the link source page 91 on the browser operation screen 90 is displayed. A sub-screen 94 is displayed on the sub-screen 94, and a summary sentence 95 of the linked page is displayed on the sub-screen 94. Of course, as shown in FIG. 17, a dedicated area 96 for displaying the summary text on the operation screen 90 of the browser may be provided.

  In general, when the browser detects that a hyperlink is described by hovering over a link location, the browser changes the shape of the mouse cursor to, for example, an icon of a hand pointing a finger from an arrow. In the link destination summary sentence presenting function, when the browser recognizes that the link location has been detected, the browser causes the browser to acquire the link destination page based on the URL describing the hyperlink set in the location. At this time, the abstract link is extracted without displaying the acquired link destination page. Specifically, in the link destination page, a character string delimited by a punctuation mark, a punctuation mark, or a line feed is extracted as a sentence, and the number of characters of each extracted sentence is counted. If the number of characters is equal to or greater than the predetermined number, morphological analysis is performed with the sentence as an analysis target sentence. In other words, by using only long sentences as candidates for summary sentences, for example, it is possible to eliminate sentences that are inappropriate as summary sentences for linked pages, such as advertisement texts included in linked pages and simple linked page descriptions. Yes. There is also an effect of reducing the processing load for extracting the summary sentence on the PC.

  Next, a morphological analysis is performed on a sentence as a summary sentence candidate as an analysis target sentence, and words included in all the analysis target sentences are extracted. Then, in all the extracted words, a word with a predetermined part of speech is used as a counting target, and the number of appearances of the counting target word is counted. In the present embodiment, the number of appearances is counted using nouns as count target words, and the number of appearances for each of all nouns is acquired as a word-specific point.

  Further, for one analysis target sentence, a total point obtained by adding the word-specific points of each noun included in the sentence is acquired as an evaluation value of the sentence. That is, each word is associated with a point by word corresponding to the number of appearances, and the point by word included in each analysis target sentence is totaled. By doing in this way, the evaluation value is made higher as the sentence includes more nouns that appear more frequently.

  For example, suppose that the number of occurrences of each noun of “Taro”, “Book”, and “Pencil” extracted from all analysis target sentences in the Web page acquired by the browser is 3, 5, and 7. Here, if there is an analysis target sentence “Taro buys a book and a pencil” in the Web page, the evaluation value of this sentence is 3 + 5 + 7 = 15 points. If evaluation values for all analysis target sentences are acquired in this way, the sentence having the highest evaluation value is finally extracted as a summary sentence, and the summary sentence is appropriately displayed. Of course, the part of speech of the word for which the number of appearances is counted is not limited to a noun, and may be another part of speech. You may comprise so that the user can set the part of speech used as a count object.

  The link destination summary presentation function may also be realized by distributed processing by the server / client system. In this case, the client computer may send a URL describing the hyperlink set at the position where the mouse is over to the server computer. The server computer acquires a Web page based on the received URL, extracts a summary sentence from the document in the page, and sends the summary sentence to the client computer. The client computer may display the sent summary sentence as appropriate.

10 sentences 11 words 12 parts of speech 16 clause boundary determination tables 18, 18b, 18c clauses 30, 80, 90 browser screen 31 information search support program operation area 32 information search support program display screen 40a Web page document 40b link reset document

Claims (24)

A program installed on a computer,
Accepts user input indicating any character description position of the acquired document, extracts characters before and after the character at the character description position in character string units based on punctuation marks or reading marks, and describes the extracted character string A keyword extraction range setting step for setting a location as a keyword extraction range;
Morphological analysis of the character string of the keyword extraction range as a processing target, decomposed in units of words, and for each decomposed word, morpheme analysis step that associates and manages each part of speech and usage as grammatical information;
A character string unit recognition step for recognizing each character string unit of a word, a phrase, and a sentence from the character string to be processed based on the grammatical information;
In accordance with user input designating a character string unit, a keyword setting step for classifying the character string to be processed for each designated character string unit, and setting a character string described by the divided character string unit as a search keyword;
A keyword inquiry step of accepting a predetermined user input, inquiring the set search keyword to a predetermined database, and displaying and outputting an inquiry result returned from the database;
An information retrieval support program characterized by causing   2. The information search support program according to claim 1, wherein in the keyword extraction range setting step, a character string range of the keyword extraction range is set according to the number of times a predetermined user input is continuously received within a predetermined time. . A program installed on a computer,
A morphological analysis step for analyzing a morphological analysis of a document acquired as a processing target and disassembling it into word units, and associating and managing each part of speech and usage as grammatical information for each decomposed word;
A character string unit recognition step for recognizing each character string unit of a word, a phrase, and a sentence from a continuous character string in a document based on the grammatical information;
A link target setting step of classifying the document into character string units designated by user input, and displaying the segmented character string unit so that it can be identified as a link target location;
A keyword setting step of setting a character string included in the link target portion as a search keyword;
A link setting step for setting a hyperlink indicating that the search keyword is referred to a predetermined database in the link target portion;
When a hyperlink set in a certain link target location is instructed by user input, a query result presentation step for displaying and outputting a query result returned from the predetermined database based on the hyperlink;
An information retrieval support program characterized by causing In claim 3,
Including a syntax analysis step of parsing the document to extract a sentence, obtaining and managing a dependency relationship in the sentence,
In the link target setting step, in the sentence, when the preceding and following clauses are in a dependency relationship, the description location of the character string connecting the clauses in the dependency relationship is set as the link target location,
In the keyword setting step, in the sentence, a character string having a dependency relation designated by user input is set as a search keyword.
An information retrieval support program characterized by that.   5. The information search support program according to claim 1, wherein, in the keyword setting step, a character string obtained by converting the phrase into a final form is set as a search keyword.   6. The occurrence frequency acquisition step for each word for obtaining the appearance frequency of each word included in the document according to claim 1, and the importance of referring to the database as a search keyword for a predetermined number of words in descending order of appearance frequency by user input. An information retrieval support program comprising a word inquiry step.   6. The information search support program according to claim 1, further comprising a keyword part-of-speech setting step for setting a specific part of speech as the search keyword by a user input.   8. The information search support program according to claim 1, wherein in the keyword setting step, a character string composed of continuous nouns is set as a compound noun, and the compound noun is set as one search keyword. .   9. The information search support program according to claim 8, wherein, in the keyword setting step, the compound noun is set by continuing nouns in a number specified by a user input. A client computer in a server-client system,
Means for displaying the acquired document;
Means for communicating with a server computer for executing morphological analysis processing;
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting means for extracting a character string unit and setting a description location of the extracted character string as a keyword extraction range;
An analysis request message sending means for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
When a message containing a search keyword is returned from the server computer that has received the analysis request message, a search result presenting means for querying the search keyword to a predetermined database and displaying the query result;
A computer having an information search support function characterized by comprising: A client computer in a server-client system,
Means for displaying the acquired document;
Means for communicating with a server computer for executing morphological analysis processing;
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting means for extracting a character string unit and setting a description location of the extracted character string as a keyword extraction range;
An analysis request message sending means for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
Search result presentation means for displaying and outputting a search keyword query result returned from the server computer that has received the analysis request message;
A computer having an information search support function characterized by comprising:   12. The information search according to claim 10, wherein the keyword extraction range setting means sets a character string range of the keyword extraction range according to the number of times a predetermined user input is continuously received within a predetermined time. Computer with support function. A server computer that communicates a computer having the information search support function according to claim 10 as a client computer,
The character string contained in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and for each decomposed word, each part of speech and usage form are associated with each other as grammatical information and managed. Morphological analysis means to perform,
Character string unit recognition means for recognizing each character string unit of words, phrases, and sentences from the character string based on the grammatical information;
Keyword setting means for dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
A search keyword inquiry message sending means for returning a keyword inquiry message including the search keyword and an instruction for making the keyword to be inquired to a predetermined database to the client computer. Server computer with support function. A server computer that communicates a computer having the information search support function according to claim 11 as a client computer,
The character string contained in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and for each decomposed word, each part of speech and usage form are associated with each other as grammatical information and managed. Morphological analysis means to perform,
Character string unit recognition means for recognizing each character string unit of words, phrases, and sentences from the character string based on the grammatical information;
Keyword setting means for dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
A server computer having an information search support function, comprising: inquiry result return means for inquiring the search keyword into a predetermined database and returning the inquiry result to the client computer. It is installed in a computer that has a display function for the acquired document and a communication function with a server computer that executes morphological analysis processing.
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting step for extracting a character string unit and setting a character string description portion of the extracted sentence as a keyword extraction range;
An analysis request message sending step for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
When a message including a search keyword is returned from the server computer that has received the analysis request message, a search result presenting step for inquiring the search keyword to a predetermined database and displaying the inquiry result;
An information retrieval support program characterized by causing It is installed in a computer that has a display function for the acquired document and a communication function with a server computer that executes morphological analysis processing.
Based on a predetermined user input, it accepts designation of a character string unit and an instruction of an arbitrary character description position in the displayed document, and the characters before and after the character at the indicated character description position are based on a punctuation mark or a punctuation mark A keyword extraction range setting step for extracting a character string unit and setting a character string description portion of the extracted sentence as a keyword extraction range;
An analysis request message sending step for sending an analysis request message including the designated character string unit, the character string of the keyword extraction range, and the designated character description position to the server computer;
A search result presentation step of displaying and outputting a search keyword query result returned from the server computer that has received the analysis request message;
An information retrieval support program characterized by causing   17. The information search according to claim 15, wherein in the keyword extraction range setting step, a character string range of the keyword extraction range is set according to the number of times a predetermined user input is continuously received within a predetermined time. Support program. Installed on a server computer that can communicate with a client computer accessed over a network,
The character string included in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and the parts of speech and usage forms of each decomposed word are managed in association with grammatical information. A morphological analysis step;
Based on the grammatical information, from the character string, a character string unit recognition step for recognizing each character string unit of a word, a clause, and a sentence;
Keyword setting step of dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
A search keyword inquiry message sending step that includes the search keyword and returns a keyword inquiry message including an instruction to make the keyword inquired to a predetermined database to the client computer;
An information retrieval support program characterized by causing Installed on a server computer that can communicate with a client computer accessed over a network,
The character string included in the analysis request message sent from the client computer is morphologically analyzed and decomposed into word units, and the parts of speech and usage forms of each decomposed word are managed in association with grammatical information. A morphological analysis step;
Based on the grammatical information, from the character string, a character string unit recognition step for recognizing each character string unit of a word, a clause, and a sentence;
Keyword setting step of dividing the character string into designated character string units according to the designation information in character string units included in the analysis request message, and setting a character string described by the divided character string units as a search keyword When,
A query result returning step of querying the search keyword to a predetermined database and returning the query result to the client computer;
An information retrieval support program characterized by causing Installed on a computer with a browser installed,
Accepts user input indicating the display position of a hyperlink to another web page set in the web page acquired by the browser, and executes a process of displaying and outputting the description content of the other web page as a summary sentence An information retrieval support program that
A linked document acquisition step of acquiring a document included in another Web page without displaying and outputting the document to a user;
An analysis target sentence specifying step for recognizing a character string unit based on a punctuation mark, a punctuation mark, or a line break as a sentence in the acquired document, and specifying a sentence including a predetermined number of characters or more as an analysis target sentence for the recognized sentence When,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, a morphological analysis step of managing the corresponding part of speech and usage as grammatical information,
Based on the grammatical information, extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words, and a word appearance number acquiring step of acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. An evaluation value acquisition step for acquiring
A summary sentence extraction step for extracting an analysis target sentence corresponding to the highest evaluation value as a summary sentence;
An information retrieval support program that executes a summary sentence display step of displaying and outputting the summary sentence extracted in the summary sentence extraction step in a display area different from the Web page document acquired by the browser. It is installed on a server computer that can communicate with a browser accessed via the network,
Acquires a Web page according to the URL sent from the browser, recognizes a character string unit based on a punctuation mark, a punctuation mark, or a line break in a document included in the Web page as a sentence. An analysis target sentence specifying step for specifying a sentence including the character as an analysis target sentence;
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, a morphological analysis step of managing the corresponding part of speech and usage as grammatical information,
Based on the grammatical information, extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words, and a word appearance number acquiring step of acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. An evaluation value acquisition step for acquiring
A summary sentence extraction step for extracting an analysis target sentence corresponding to the highest evaluation value as a summary sentence;
A summary sentence returning step of returning the summary sentence extracted by the summary sentence extracting step to the browser;
An information retrieval support program that executes Implements a browser, accepts user input indicating the display position of a hyperlink to another web page set in the web page acquired by the browser, and makes the description content of the other web page a summary sentence A computer for display output,
A linked document acquisition means for acquiring a document contained in another Web page without displaying and outputting it to a user;
In the acquired document, a character string unit based on a punctuation mark, a punctuation mark, or a line feed is recognized as a sentence, and a sentence including a predetermined number of characters or more is identified as an analysis target sentence for the recognized sentence. When,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, morphological analysis means that associates and manages each part of speech and usage as grammatical information,
Based on the grammatical information, the word appearance frequency acquisition means for extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words and acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. Evaluation value acquisition means for acquiring
Summary sentence extraction means for extracting the analysis target sentence corresponding to the highest evaluation value as a summary sentence;
Summary text display means for displaying and outputting the summary text extracted by the summary text extraction means in a display area different from the document of the Web page acquired by the browser;
A computer having an information search support function characterized by comprising: A server computer in a server-client system,
Means for communicating with a browser accessed over a network;
Acquires a Web page according to the URL sent from the browser, recognizes a character string unit based on a punctuation mark, a punctuation mark, or a line break in a document included in the Web page as a sentence. An analysis target sentence specifying means for specifying a sentence including a character as an analysis target sentence,
Morphological analysis of the identified sentence to be analyzed and decomposed into units of words, and for each decomposed word, morphological analysis means that associates and manages each part of speech and usage as grammatical information,
Based on the grammatical information, the word appearance frequency acquisition means for extracting words of a predetermined part of speech included in all the analysis target sentences as counting target words and acquiring the number of appearances of each counting target word;
The number of occurrences of each count target word is used as a point for each word, and the total value of each point for each count target word included in one analysis target sentence is used as the evaluation value for that sentence. Evaluation value acquisition means for acquiring
Summary sentence extraction means for extracting the analysis target sentence corresponding to the highest evaluation value as a summary sentence;
Summary sentence return means for returning the summary sentence extracted by the summary sentence extraction means to the browser;
A server computer having an information retrieval support function.   The program storage medium which memorize | stored the information search assistance program in any one of Claims 1-9 or any one of Claims 15-21.

Images