JP2015095180A - Content search program for clearly presenting selection candidates for query word in refined search, device, and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a content search program and the like which clearly present perspectives for an input query word to a user as selection candidates, and perform a further refined search by the selected perspective.SOLUTION: A device comprises: content text collection means which can store a description of each content or search for a description of each content via a network; content search means which searches for a plurality of contents in descending order of the number of words corresponding to the input query word; declinable word/indeclinable word extraction means which extracts a plurality of declinable words or indeclinable words from the plurality of descriptions of the plurality of retrieved contents; search result clear presentation means which clearly presents description of each content to a user, and requests the user to select one or more declinable words or indeclinable words from the plurality of extracted declinable words or indeclinable words as next search keys; and re-search input means which inputs the declinable words or indeclinable words selected by the user as query words of the next search keys, in the content search means.

Description

  The present invention relates to a search engine technique for searching for content based on an input keyword.

  According to a general search engine such as Google (registered trademark) or Yahoo! (registered trademark), search results of a plurality of contents are specified based on a user input keyword. At this time, if there is a possibility that the input keyword has a spelling error, there is a technique for clearly indicating the corrected keyword to the user. Further, not only the input keyword but also other candidate keywords can be combined to specify a list of keywords that are frequently input.

  2. Description of the Related Art Conventionally, there is a technique for searching for content based on a user input keyword and clearly indicating to the user a “related keyword” that is optimal for the search from the keyword (see, for example, Patent Document 1). By causing the user to select these related keywords, it is possible to further narrow down the search results and to perform an information search with less noise with an improved hit rate. According to this technique, for example, for the input keyword “Osaka”, related keywords such as “Osaka City” and “Osaka Waterworks Bureau” related to “Osaka” are clearly indicated to the user. The user can further narrow down and search by selecting any related keyword.

JP 2006-163645 A

  According to the prior art, the user's input keyword itself is corrected, or other related keywords that complement the input keyword are derived.

  Here, for example, a case is assumed where an application of “simple” “household account book” is searched. For example, assume that the user wants to search for an application for an easy-to-input household account book. As the simple input operation, for example, a selection type (tap type) operation is assumed, and keywords such as “selection type” and “household account book” may be preferably used as query words.

  However, it is difficult for the user to set the query word by specifically imagining the function of the application like “selection formula”. Actually, there are cases where you don't know the existence of such a function at all, or you just want to search with an ambiguous intention such as "an app that is not difficult anyway". Even in such a case, it is desirable to be able to search by “easy” and “household account book”.

  On the other hand, there are a large number of applications described as “simple”, and there are various types of applications as “input” or “display” as a viewpoint called “simple”. As a user, it is necessary to check the viewpoint referred to as “easy” while reading each explanatory sentence from a large number of search results searched by the query words “easy” and “household account book”.

  According to the technique described in Patent Document 1 described above, “easy operation” and “simple input” can be recommended to the user as words including a query word. However, “Very easy to operate” and “Easy to input with selection” are not recommended because they are not words. Further, according to an existing search engine, it is possible to recommend a combination of query words input in the past by a large number of users, such as “Easy household account book”. However, it does not necessarily meet the user's requirements. In fact, when “Easy household account book” was entered into an existing search engine, “Easy household account book app” and “Easy household account book easy” were recommended. Here, keyword candidates representing the “easy” viewpoint are not recommended. If only “Household account book” is entered, “Household account book free software” and “Household account book app” are recommended.

  In addition, as a general related word extraction technique, there is a technique for extracting a plurality of words using as an index the high possibility of co-occurring in the same sentence. However, there also arises a problem that words with high appearance frequency such as “application” and “free” are easily extracted. In addition, since the relationship between the words is not taken into account, the irrelevant words at a distant position may be extracted.

  Accordingly, an object of the present invention is to provide a content search program, an apparatus, and a method for clarifying a viewpoint of an input query word as a selection candidate to a user and further narrowing down a search based on the selected viewpoint.

According to the present invention, in a content search program for executing a computer mounted on an apparatus,
Content text collecting means capable of accumulating or searching through the network explanatory text relating to the content for each content;
Content search means for searching for a plurality of contents in the order including many words that match the input query word using content information collection means;
A prescriptive word extracting means for extracting a plurality of remarks or expressions from a plurality of explanatory texts of a plurality of contents searched by the content search means;
The explanatory text searched by the content search means is clearly indicated to the user for each content, and any one or more of the words or phrases are extracted from the plurality of extracted words or phrases as the next search key. Search result explicit means to be selected,
The computer is caused to function as a re-search input unit that inputs a prey or body selected by the user as a query word of the next search key to the content search unit.

According to another embodiment of the content search program of the present invention,
Predicates are adjectives, adjective verbs and / or verbs as inflections
Body language is a noun as an independent word,
It is also preferred that the computer function so that the related word is a particle as a dependency word.

According to another embodiment of the content search program of the present invention,
By the re-search input means, the prescriptive word extraction means and the search result clarification means are repeated at least twice,
The prescriptive word extraction means first extracts a prescriptive word (or prescriptive word), and then extracts a prescriptive word (or prescriptive word),
The search result clarification means first causes the user to select a utterance (or statement), and then causes the user to select a utterance (or statement).
The re-search input means inputs the word (or expression) selected by the user first to the content search means, and then inputs the expression (or expression) selected by the user to the content search means. It is also preferable to make it function.

According to another embodiment of the content search program of the present invention,
A word pair registration means for extracting and registering word pairs composed of two words, a predicate and a body word connected by related words;
It is also preferable that the prescriptive word extracting means causes the computer to function so as to extract a plurality of predicates or words from the explanatory text using the word pair registering means.

According to another embodiment of the content search program of the present invention,
In order to register word pairs in the word pair registration means,
A related word registration means for registering at least one related word;
For each content, a word pair of a predicate or a body language that is modified by a related word registered in the related word registration unit is searched from the content information collecting unit, and the word pair is acquired by registering the word pair in the word pair registration unit. Means,
Using the word pair registered in the word pair registration means, the related information for the word pair is searched from the content information collecting means, and the computer further functions as a related word acquisition means for registering the related word in the related word registration means It is also preferable that

According to another embodiment of the content search program of the present invention,
In the content search device,
Content text collecting means capable of accumulating or searching through the network explanatory text relating to the content for each content;
Content search means for searching for a plurality of contents in the order including many words that match the input query word using content information collection means;
A prescriptive word extracting means for extracting a plurality of remarks or expressions from a plurality of explanatory texts of a plurality of contents searched by the content search means;
The explanatory text searched by the content search means is clearly indicated to the user for each content, and any one or more of the words or phrases are extracted from the plurality of extracted words or phrases as the next search key. Search result explicit means to be selected,
It is also preferable to have a re-search input means for inputting the prey or body selected by the user as a query word for the next search key to the content search means.

According to the present invention, in a content search method using an apparatus,
For each content, it has a content text collection unit that can store explanatory text related to the content or search via the network,
A first step of searching for a plurality of contents in an order including a large number of words that match the input query word using the content information collection unit;
A second step of extracting a plurality of remarks or body words from a plurality of explanatory texts of a plurality of contents searched in the first step;
The explanatory texts of the plurality of contents searched in the first step are clearly shown to the user, and any one or more words or phrases are extracted from the plurality of extracted words or phrases as the next search key. A third step for the user to select;
A fourth step of using the content information collection unit to search for a plurality of contents in the order that includes a large number of matching words, using the word or phrase selected by the user as the query word of the next search key;
And a fifth step of clearly explaining to the user explanatory texts of the plurality of contents searched in the first step.

According to another embodiment of the content search method of the present invention,
As a next search key, a sixth step for allowing the user to select any one of the extracted words or phrases from the plurality of extracted words or phrases;
A seventh step of searching for a plurality of contents in the order including a large number of matching words, using the content information collecting unit as a query word of the next search key with the body word or prey selected by the user;
It is also preferable to have an eighth step of clearly explaining to the user explanatory texts of the plurality of contents searched in the seventh step.

According to another embodiment of the content search method of the present invention,
A word pair registration unit that extracts and registers word pairs composed of two words, a predicate and a body word, connected by related words;
In the second step, it is also preferable to extract a plurality of predicates or body words from the explanatory text using the word pair registration unit.

According to another embodiment of the content search method of the present invention,
In order to register word pairs in the word pair registration unit,
A related word registration unit for registering at least one related word;
For each content, searching the content information collecting means for a word pair of a pretext or body language that is dependent on the related word registered in the related word registration means, and registering the word pair in the word pair registration means;
It is preferable that the method further includes a step of retrieving a related word for the word pair from the content information collecting unit using the word pair registered in the word pair registering unit and registering the related word in the related word registering unit.

  According to the content search program, apparatus, and method of the present invention, it is possible to clearly specify the viewpoint of the input query word as a selection candidate to the user, and to further narrow down the search from the selected viewpoint. Therefore, even if a user inputs a sensuous and ambiguous query word, the user can immediately select a desired keyword to be narrowed down by looking at a keyword selection candidate as a viewpoint.

It is a functional block diagram of the content search apparatus in this invention. It is a flowchart of the content search method in this invention. It is a screen display figure of the terminal in S11 of FIG. It is a screen display figure of the terminal in S14 of FIG. It is a screen display figure of the terminal in S21 of FIG. It is a screen display figure of the terminal in S24 of FIG. It is a screen display figure of the terminal in S31 of FIG. It is a screen display figure of the terminal in S33 of FIG. It is explanatory drawing showing the registration information of the word pair registration part in this invention, and a related word registration part.

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

  FIG. 1 is a functional configuration diagram of a content search apparatus according to the present invention.

  The content search apparatus 1 receives a query word (search word) from the terminal 2 operated by the user, and searches for a plurality of applications (contents) using the query word as a key. The content search apparatus 1 acquires and accumulates application names, explanatory texts, and acquisition URLs (Uniform Resource Locators) from a large number of server groups 3 in advance. The terminal 2 is preferably a terminal such as a smartphone or a tablet. According to the content search apparatus 1 of the present invention, since the keyword of the viewpoint is displayed as a selection candidate for the input query word, the user can further narrow down and search by simply tapping it.

  According to FIG. 1, the content search apparatus 1 can be broadly divided into <content search function> and <word pair collection function>. The <content search function> includes a content information collection unit 101, a content search unit 111, a noun phrase extraction unit 112, a search result clarification unit 113, and a re-search input unit 114. The <word pair collection function> further includes a word pair registration unit 102, a related word registration unit 103, a word pair acquisition unit 121, and a related word acquisition unit 122. These functional components can be realized by executing a program that causes a computer installed in the apparatus to function.

<Content search function>
The content search function searches for an application (content) based on the query word received from the terminal 2 of the user, and returns the viewpoint for the query word as a selection candidate together with the search result.

[Content Information Collection Unit 101]
The content information collection unit 101 is a database in which a large amount of text (explanatory text) related to the content is accumulated for each content. Alternatively, the content information collection unit 101 may be capable of searching for content information stored in various server groups connected to the Internet via a network. The content information collection unit 101 is referred to by the content search unit 111.

  The content is, for example, an application, and the user can search for a desired application. The explanatory text is text describing the application, not the application itself. This is only for reference when the user searches for a desired application.

[Content Search Unit 111]
The content search unit 111 receives a query word from the terminal 2. Then, the content information collection unit 101 is used to search for a plurality of contents in the order including many words that match the query word. The searched explanation text of the content is output to the prescriptive word extraction unit 112. Note that query words may be extracted from the query text input by the user from the terminal 2 by morphological analysis.

[Phrase word extraction unit 112]
The prescriptive word extraction unit 112 extracts a plurality of predicates or a word from a plurality of explanatory texts of a plurality of contents searched by the content search unit 111.
“Predicate” is an adjective, an adjective verb and / or a verb as an independence word.
A “body word” is a noun as a word independent word.
Here, the prescriptive word extracting unit 112 can also extract a plurality of prescriptive words or sample words from the explanatory text by, for example, the following two methods.
-Extract a predicate or body word registered in the word pair registration unit 102.
・ Extract prescriptive or body words by morphological analysis of explanatory texts.
The extracted prescription or body phrase is output to the search result clarification unit 113.

  Moreover, it is preferable that the prescriptive word extraction unit 112 is repeated at least twice or more by the re-search input unit 114 described later. In this case, the prescriptive word is extracted first, and then the word is extracted. On the contrary, it is also possible to extract the body first and then extract the pretext.

[Search Result Clarifying Unit 113]
The search result clarification unit 113 clarifies the explanatory text searched by the content search unit 111 to the user for each content, and uses any one or more of the extracted remarks or body words as the next search key. Let the user select a precaution or body phrase.

  Moreover, it is preferable that the search result clarification part 113 is repeated at least twice or more together with the prescriptive phrase extraction part 112 by the re-search input part 114 described later. In this case, the user first selects a word, and then causes the user to select a word. On the contrary, you may make it select a word first and then select a word.

[Re-search input unit 114]
The re-search input unit 114 inputs the word or body selected by the user to the content search unit 111 as a query word for the next search key. In particular, the re-search input unit 114 preferably repeats the prescriptive phrase extraction unit 112 and the search result clarification unit 113 at least twice. That is, after searching for an application based on a query word input by a user, the search results are sorted according to a combination of prescriptive and body words.

  According to the present invention, the user who operates the terminal 2 can display the search result based on the viewpoint only by tapping the search candidate keyword displayed automatically after inputting the query word. Therefore, the terminal 2 is suitable for a device equipped with a touch panel display such as a smartphone or a tablet.

  FIG. 2 is a flowchart of the content search method according to the present invention. In addition, according to FIG. 2, it can understand as a content search method similarly to the flow of a process of the function structure part of FIG.

(S11) FIG. 3 is a screen display image of the terminal in S11 of FIG.
It is assumed that the user desires to install the “household account book” application (content) on the terminal 2. The terminal 2 accesses the content search device 1 and displays the search page. At this time, it is assumed that “household account book” is input to the query input field by the user (see FIG. 3). The terminal 2 transmits a search request including this query word to the content search device 1.

(S12) The content search device 1 uses the content information collection unit 101 to search for a plurality of applications in the order including many words that match the input query word “household account book”. For example, it is assumed that the following application that includes many query words “household account book” in the application description is searched.
<Application name><Descriptiontext>
"Easy household account book" (Easy to save household account book app ...)
"Kake ~ bo" (Easy input! Simple input! ...)
"Free and household account book app" (a cute design dog ... very simple input)
“Money Management” (Household account book application with multi-functional operation! Cute calculator ...)
・ ・ ・ ・ ・ ・

(S13) The content search device 1 first extracts a plurality of “uses” from a plurality of explanatory texts of the plurality of contents searched in S12. A noun is an adjective, an adjective verb and / or a verb. Assume that the following terms are searched in the order of appearance frequency.
"Easy""Simple""Convenient""Highfunctionality""Cute""Busy"

(S14) FIG. 4 is a screen display image of the terminal in S14 of FIG.
The content search device 1 describes the description texts of the plurality of searched applications and the plurality of “use words” extracted as the next search key on the page. The page is returned to the terminal 2 as a search response. The terminal 2 displays the page received from the content search device 1 on the display and makes the user recognize it.

(S21) FIG. 5 is a screen display image of the terminal in S21 of FIG.
The terminal 2 causes the user to select a precaution to be narrowed down as a search key next. At this time, it is assumed that the user has checked “simple” and “cute”. The terminal transmits a search request including the phrases “simple” and “cute” to the content search device 1.

(S22) The content search apparatus 1 further narrows down and searches using the words “simple” and “cute” as search keys.

(S23) Next, the content search device 1 extracts a plurality of “body words” from a plurality of explanatory texts of a plurality of searched contents. A “body word” is a noun. Assume that the following phrases are searched in the order of appearance frequency.
The word "simple"-> the word "display""input"
Precautionary word "cute"-> words "design""calculator"
In this way, even if it is a sensual query word such as “simple” or “cute”, the user can recognize at a glance a keyword that is a viewpoint that embodies it.

(S24) FIG. 6 is a screen display image of the terminal in S24 of FIG.
The content search device 1 describes the description texts of the plurality of searched applications and the plurality of “body words” extracted as the next search keys on the page. The page is returned to the terminal 2 as a search response. The terminal 2 displays the page received from the content search device 1 on the display and makes the user recognize it.

(S31) FIG. 7 is a screen display image of the terminal in S31 of FIG.
The terminal 2 causes the user to select a statement to be narrowed down next as a search key. At this time, it is assumed that the user has checked the utterance “input” with respect to the predicate “simple”. The terminal transmits a search request including the word “input” to the content search device 1.

(S32) The content search device 1 further narrows down and searches using the word “input” + the word “simple” as a search key.

(S33) FIG. 8 is a screen display image of the terminal in S33 of FIG.
Finally, the user can search intuitively without knowing the function name of the application.

<Word collection function>
The word pair collection function is for learningly registering “word pairs” and “related words” in the word pair registration unit 102 described above.

  “Word pair” refers to a pair of two words, a predicate and a body word, connected by related words. A “word” is one of the structural units of a language and consists of one or more morphemes. A morpheme is the smallest unit that has meaning for a language, and refers to each of a set of phonemes that have been decomposed into units that no longer make sense after further decomposition. The word includes a simple word (for example, a mountain) composed of one morpheme and a compound word (for example, a mountain climb) composed of a plurality of morphemes. In the following, when referring to “words”, both simple words and compound words are targeted.

  A “related word” is a function word having a grammatical role and means a case particle as a dependency word. For example, there are "... is ...", "... is ...", "... is ...".

  FIG. 9 is an explanatory diagram showing registration information of the word pair registration unit and the related word registration unit in the present invention.

[Word Pair Registration Unit 102]
The word pair registration unit 102 registers “word pairs”. The subsequent registration method will be described later by the word pair acquisition unit 121.

[Related Word Registration Unit 103]
The related word registration unit 103 registers at least one related word. The subsequent registration method will be described later by the related word acquisition unit 122.

[Word Pair Acquisition Unit 121]
For each content, the word pair acquisition unit 121 searches the content information collection unit 101 for “word pairs” of the predicates and body words that are modified by the “related words” registered in the related word registration unit 103, and The word pair is registered in the word pair registration unit 102.

For example, first, consider a case where the case particle “NA” is registered manually in the related word registration unit 103. In this case, it is assumed that there is content in which the following explanatory text is described.
(Example) "... Simple input ..."
<Simple (prescriptive)> Na <Input (body)>
In this case, “simple” and “input” are registered in the related word registration unit 103 as word pairs.

[Related Word Acquisition Unit 122]
Using the word pair registered in the word pair registration unit 102, the related word acquisition unit 122 searches the content information collection unit 101 for a related word for the word pair, and registers the related word in the related word registration unit 103. To do.

The word pairs “simple” and “input” acquired in the example of the word pair acquisition unit 121 described above are used. The related word acquisition unit 122 searches the content information collection unit 101 using a search key in which a wild card * is inserted between word pairs.
<Input> * <Simple>
<Simple> * <Input>
In this case, it is assumed that there is content in which the following explanatory text is described.
(Example) "... Input is simple ..."
Input (G) Simple The related word acquisition unit 122 registers the related word “ga” in the related word registration unit 103.
Pattern: [<Prescriptive / Prescriptive> (Related terms) <Prescriptive / Prescriptive>]]
<Symposium> is <Use>
Thus, a new relation pattern can be extracted by the relation word “ga”.

By alternately repeating the processing of the word pair acquisition unit 121 and the related word acquisition unit 122, the pattern of the word pair registration unit 102 (and the related word registration unit 103) can be automatically learned and accumulated. . In this case, it is assumed that there is content in which the following explanatory text is described.
(Example 2) "... Easy to enter ..."
<Input (word)> is <simple (word)>
In this case, “input” and “simple” are registered in the related word registration unit 103 as word pairs.

It should be noted that “related words” are intended only for those having an appearance frequency of a predetermined number or more over the entire content information in the content. Specifically, the top 10 items may be used.
For example, if there is a sentence “Easy operation for input”, the related word is “Operation for”. However, since this related word hardly appears, it is not registered in the related word registration unit 103 as a related word.

Here, the “appearance frequency” of the word pair (word A, word B) and the related word C will be described. It is assumed that the query word A is extracted from the query word (sentence for search) input by the user.
First, content information is searched from the content information collection unit 101 in the order of appearance frequency of the query word A (see the content search unit 111 in FIG. 1 and S23 in FIG. 2). For each piece of content information, the word pair (word A, word B) and the appearance frequency a of the related word C are calculated.
Next, content information is retrieved from the content information collection unit 101 in the order of application frequency of the word B. Then, for each content information, the word pair (word B, word D) and the appearance frequency b of the related word E are calculated.
Finally, the total value of the appearance frequencies a and b may be the “appearance frequency” of the word pair (word A, word B) and the related word C.

  As described above in detail, according to the content search program, apparatus, and method of the present invention, the viewpoint for the input query word is clearly indicated as a selection candidate to the user, and the search is further narrowed down based on the selected viewpoint. be able to. Therefore, even if a user inputs a sensuous and ambiguous query word, the user can immediately select a desired keyword to be narrowed down by looking at a keyword selection candidate as a viewpoint. In particular, by presenting a predicate such as an adjective first, it is easy for the user to narrow down sensuously.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Content search device 101 Content information collection part 111 Content search part 112 Phrase extraction part 113 Search result clarification part 114 Re-search input part 102 Word pair registration part 103 Related word registration part 121 Word pair acquisition part 122 Relation word acquisition part 2 Terminal 3 server group

Claims (10)

In a content search program for executing a computer mounted on an apparatus,
Content text collecting means capable of accumulating or searching through the network explanatory text relating to the content for each content;
Content search means for searching for a plurality of contents in order including many words that match the input query word using the content information collection means;
A prescriptive word extracting means for extracting a plurality of remarks or words from a plurality of explanatory texts of a plurality of contents searched by the content search means;
The explanatory text searched by the content search means is clearly indicated to the user for each content, and as the next search key, one or more of the extracted words or phrases are extracted from the extracted words or phrases. Search result clarification means to be selected,
A content search program for causing a computer to function as a re-search input means for inputting a prey or body word selected by a user as a query word of a next search key to the content search means. The predicates are adjectives, adjective verbs and / or verbs as independence words,
The body language is a noun as a word independent word,
The content search program according to claim 1, wherein the computer is caused to function so that the related word is a particle as a dependency word. By the re-search input means, the prescriptive word extraction means and the search result clarification means are repeated at least twice,
The prescriptive word extracting means first extracts a prescriptive word (or a prescriptive word), and then extracts a prescriptive word (or prescriptive word),
The search result clarification means first causes the user to select a pretext (or a word), and then causes the user to select a word (or a word),
The re-search input means inputs a word (or a body word) first selected by the user to the content search means, and then inputs a word (or body word) selected by the user to the content search means. The content search program according to claim 1, wherein the computer is caused to function. A word pair registration means for extracting and registering word pairs composed of two words, a predicate and a body word connected by related words;
4. The prescriptive / phrase extraction unit causes the computer to function so as to extract a plurality of predicates or dialects from the explanatory text using the word pair registration unit. The content search program described in the section. In order to register word pairs in the word pair registration means,
A related word registration means for registering at least one related word;
For each content, a word pair of a predicate or a body language that is modified by a related word registered in the related word registration unit is searched from the content information collection unit, and the word pair is registered in the word pair registration unit. Word pair acquisition means;
Using the word pair registered in the word pair registration unit as a related word acquisition unit for searching the related information for the word pair from the content information collecting unit and registering the related word in the related word registration unit The content search program according to claim 4, further causing a computer to function. In the content search device,
Content text collecting means capable of accumulating or searching through the network explanatory text relating to the content for each content;
Content search means for searching for a plurality of contents in order including many words that match the input query word using the content information collection means;
A prescriptive word extracting means for extracting a plurality of remarks or words from a plurality of explanatory texts of a plurality of contents searched by the content search means;
The explanatory text searched by the content search means is clearly indicated to the user for each content, and as the next search key, one or more of the extracted words or phrases are extracted from the extracted words or phrases. Search result clarification means to be selected,
A content search apparatus comprising: a re-search input unit that inputs a prey or body selected by a user as a query word of a next search key to the content search unit. In a content search method using a device,
For each content, it has a content text collection unit that can store explanatory text related to the content or search via the network,
A first step of searching for a plurality of contents in the order including many words that match the input query word using the content information collection unit;
A second step of extracting a plurality of remarks or body words from a plurality of explanatory texts of a plurality of contents searched in the first step;
The explanatory texts of the plurality of contents searched in the first step are clearly shown to the user, and any one or more words or phrases are extracted from the plurality of extracted words or phrases as the next search key. A third step for the user to select;
A fourth step of searching for a plurality of contents in the order including a large number of matching words, using the content information collecting unit as a query word of the next search key, with the word or phrase selected by the user,
And a fifth step of clearly indicating to the user explanatory texts of the plurality of contents searched in the first step. As a next search key, a sixth step for allowing the user to select any one of the extracted words or phrases from the plurality of extracted words or phrases;
A seventh step of searching for a plurality of contents in the order including many matching words, using the content information collecting unit as a query word of the next search key, with the body word or prey selected by the user;
The content search method according to claim 7, further comprising an eighth step of clearly explaining to the user explanatory texts of the plurality of contents searched in the seventh step. A word pair registration unit that extracts and registers word pairs composed of two words, a predicate and a body word, connected by related words;
9. The content search method according to claim 8, wherein in the second step, a plurality of remarks or body words are extracted from the explanatory text using the word pair registration unit. In order to register word pairs in the word pair registration unit,
A related word registration unit for registering at least one related word;
For each content, a word pair of a predicate or a body language that is modified by a related word registered in the related word registration unit is searched from the content information collection unit, and the word pair is registered in the word pair registration unit. Steps,
Using the word pair registered in the word pair registering unit, searching for a related word for the word pair from the content information collecting unit, and registering the related word in the related word registering unit. The content search method according to claim 7, wherein:

Images