JP2581376B2 - Document search device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval apparatus,
In particular, the present invention relates to a document search device having a search function for documents and their differences / similarities.

[0002]

2. Description of the Related Art A conventional document search apparatus is a document search apparatus that performs a search by inputting only the attributes of a document as search conditions, narrowing down the number of documents by a plurality of search conditions, and searching for a target document. In a document search, when a single document is searched and the document is not the target document but is felt to be close, a search request based on the similarity in terms of being very similar to the target document is issued. In some cases, a search request may be made with a difference in terms of being different in some respects that are the same as the target document in some respects. For example, Japanese Patent Application Laid-Open No. 2-2458 proposes a method of automatically extracting a keyword from a document to which no keyword is assigned, calculating the similarity between the keyword and a keyword given as a search key, and searching for the document. Have been
Even if a document that does not have a keyword is given as a search key, a similar document can be searched by extracting the keyword from the document.

The conventional apparatus does not provide a search means for searching for a similar document using the displayed document itself as a search key. Therefore, a keyword for specifying a document to be searched is imagined and the keyword is searched for. There is a problem that a search cannot be performed correctly because a keyword cannot be accurately specified for a document.

[0004] Even the search method described in the above-mentioned publication does not provide a search method in which a document is used as a search source and a difference or similarity with the document is further designated.

[0005] As described above, in the conventional document retrieval apparatus, when one document is retrieved or already presented, it is difficult to specify a document retrieval similar to the document or to obtain a certain viewpoint with the document. There is a problem that a search for a different document cannot be specified.

[0006]

In order to solve the above-mentioned problems, a document search apparatus according to the present invention comprises: a document holding unit for holding a document; a document attribute holding unit for holding an attribute attached to the document; Search condition input means for inputting a search condition of a document, document search means for searching for a document according to the search condition, document presentation means for presenting the search document searched by the document search means and attribute information related to the search document Document attribute value condition input means for inputting the search document and attribute values as search conditions, document attribute type condition input means for inputting the search document and attribute types as search conditions, and presentation of neighborhood of the search document Document proximity request input means for inputting a request of the same, similarity calculation means for calculating similarity from the attribute of the search document and the attribute of another document, the attribute of the search document and another sentence A difference calculating means for calculating the difference from the attribute of the document, and the calculated similarity or difference and a search condition obtained by the document attribute value condition input means, the document attribute type condition input means and the document neighborhood request input means. A similarity / difference search means for searching for a document, and a proximity information presenting means for presenting similarity / difference information between the search document and its neighboring documents when the search result is a plurality of documents.

[0007]

Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing one embodiment of the present invention.
Referring to FIG. 1, in an embodiment of the present invention, a document database 1 including therein a document holding unit 2 for holding a document and a document attribute holding unit 3 for holding an attribute attached to the document, and a search condition of the document. Search condition input means 4 for inputting, document search means 5 for searching for a document according to the search condition input by the search condition input means 4, and document search means 5
A document presenting unit 12 for presenting the search document and attribute information related to the search document searched by the search unit; a document attribute value condition input unit 6 for inputting the search document and attribute values as search conditions; Attribute type input means 7 for inputting the type of the search document as a search condition, document neighborhood request input means 8 for inputting a request for presentation of the vicinity of the search document, and similarity from the attribute of the search document and the attribute of another document Similarity calculating means 9 for calculating the similarity, and difference calculating means 1 for calculating the difference from the attribute of the search document and the attribute of another document.
0, a similarity / difference search for searching a document in accordance with the calculated similarity or difference and the search conditions obtained by the document attribute value condition input means 6, the document attribute type condition input means 7, and the document neighborhood request input means 8 A means 11 for presenting similarity / difference information between the search document and its neighboring documents when the retrieval result is a plurality of documents;
It is composed of

[0008] The ordinary search condition of the document input by the search condition input means 4 is transmitted through the communication line 45 to the document search means 5.
Sent to In the document search means 5, the document attribute storage means 3
More document attributes are extracted through the communication line 35, attributes matching the conditions are searched, and the corresponding document is extracted from the document storage unit 2 through the communication line 25. The retrieved document is sent to the document presenting means 12 through the communication line 512 and presented to the user. The specific document (present document) and the attribute value input by the document attribute value condition input unit 6 are transmitted through a communication line 611, and the specific document and the attribute type input by the document attribute type condition input unit 7 are transmitted through a communication line 711. The distance between the specific document and the neighborhood inputted by the document neighborhood request input means 8 is sent to the similarity / difference search means 11 through the communication line 811. The similarity / difference searching means 11 retrieves the document attribute from the document attribute storing means 3 through the communication line 311. The attribute of the search document and the attribute of the display document are sent to the similarity calculating means 9 through the communication line 119 and to the difference calculating means 10 through the communication line 1110. The similarity and difference calculated by the similarity calculator 9 and the difference calculator 10 are sent to the similarity / difference calculator 11 through the communication line 911 and the communication line 1011 respectively. Similarity / difference / difference search means 1
In step 1, a document that satisfies the search condition is searched based on the obtained similarity / difference, and the corresponding document is extracted from the document storage unit 2 through the communication line 211. If one document is found, the document is sent to the document presenting means 12 through the communication line 1112 and presented to the user. If there are a plurality of retrieved documents, the retrieved documents are sent to the neighborhood information presenting unit via the communication line 1113, and the retrieved documents are presented together with the similarity / difference between the documents.

In the present embodiment, the attributes include keywords existing in the document and keywords assigned to the document.

In the document attribute value condition input means 6, the user inputs to the currently presented document that the document is different from or the same as the document. For example, the relevant presentation document was written by "Sato", but the document to be searched was specified by me, or this document was further specified by the keyword "economic friction". From the keyword "Prime Minister".

The document attribute type condition input means 7 specifies a search using an attribute type that is different or the same as the attribute when the user cannot clearly specify the difference from the currently presented document. For example, the user cannot remember when the document was written, but specifies that the document has a different "date and time" from this document.

In the document neighborhood request input means 8, when the user is conscious that the document presently presented is not this document but very similar, the user designates that the document is close in attribute to the document. .

In response to these search conditions, the similarity calculation means 9 classifies documents having more common attributes as similar documents. Further, the difference calculating means 10 further classifies differences between similar documents based on differences in attributes. As these calculation methods, for example, a method of calculating the number of different attribute values of two documents as a distance between the documents, and classifying the documents using this distance can be used. The similarity / difference search unit 11 searches for a document that satisfies the search condition using the calculated similarity and difference between the documents.

When there are a plurality of sentences in the search result, the neighborhood information presenting means 13 adds the difference between the document that was the search condition in the immediately preceding search and the searched document to the searched document. Present to the user. As a presentation method,
For example, it is possible to use a method in which the document of the search condition and the retrieved document are represented by a graph structure having nodes as nodes, and a line connecting the document of the search condition and the retrieved document is given an attribute value indicating the difference.

[0015]

As described above, in the document retrieval apparatus according to the present invention, when one document is retrieved or already presented, designation of document retrieval similar to the document, This has an effect that a document search can be performed by designating a search for a document having a different viewpoint from the document. In addition, this document search apparatus handles similarity / difference not only for a document composed of characters but also for additional attributes such as image data and audio data, so that a similarity / difference of a multimedia document can be obtained. You can also search.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 document database 2 document storage means 3 document attribute storage means 4 search condition input means 5 document search means 6 document attribute value condition input means 7 document attribute type condition input means 8 document neighborhood request input means 9 similarity calculation means 10 difference calculation Means 11 Similarity / difference search means 12 Document presentation means 13 Neighborhood information presentation means

Claims (1)

(57) [Claims] 1. A document holding unit for holding a document, a document attribute holding unit for holding an attribute attached to the document, a search condition input unit for inputting a search condition of the document, and a document search for searching the document according to the search condition Means, document presenting means for presenting the search document searched by the document search means and attribute information about the search document, document attribute value condition input means for inputting the search document and attribute values as search conditions, A document attribute type condition input unit for inputting the search document and the attribute type as search conditions, a document neighborhood request input unit for inputting a request for presentation of the vicinity of the search document; A similarity calculating means for calculating the similarity from the attribute, a difference calculating means for calculating the difference from the attribute of the search document and the attribute of another document, and a calculated similarity or difference. A similarity search for a document according to the search conditions obtained by the document attribute value condition input means, the document attribute type condition input means and the document neighborhood request input means.
A document search device comprising: a difference search unit; and, when a search result is a plurality of documents, a neighborhood information presenting unit that presents similarity / difference information between the search document and a neighboring document thereof.