JP4075094B2 - Information classification device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus for classifying a file including text composed of character codes with a keyword attached thereto.
[0002]
[Prior art]
In general, a database system provides an interface for inputting a keyword logical expression or the like so that a user having a search purpose can easily reach a target file. However, such an interface is not very useful for users who have no particular search purpose and are interested in what files are stored in the database. Conventionally, in order to provide a list of database contents, the database administrator prepares a fixed conceptual system in advance, understands the contents of the newly added file, determines the position in the system, The provider specified the location, or compared the file already classified by hand with the keyword and automatically classified it to the closest location, and provided the classification result to the user.
[0003]
[Problems to be solved by the invention]
The above-described method for preparing a fixed concept system has a problem that when a file having a new concept appears, its existence is not transmitted to the user. In automatic classification, if even one of the already classified files and a keyword match, it is determined that they are conceptually close and only classified into existing concepts. Therefore, it is necessary to correct the concept system at an appropriate time and start the classification again, but the work is enormous in proportion to the scale of the database.
[0004]
The present invention uses a keyword contained in a file as a concept, rather than using a fixed concept system, automatically classifies the file as a set belonging to it, and provides the user with a list of database contents. Objective.
[0005]
[Means for Solving the Problems]
The present invention provides a file storage means for storing a file including text composed of character codes, and a morpheme analysis for performing a morphological analysis on a text portion of the file output by the file storage means and outputting the file together with an identifier of the file A keyword storage means for storing a keyword for classifying the file, an output of the keyword storage means and an output of the morpheme analysis means, and the keyword storage means from among the results of the morpheme analysis A keyword collecting unit that collects a keyword that matches a stored keyword and outputs the keyword together with the identifier of the file, and a keyword collected by the keyword collecting unit includes a number of files including the keyword greater than a predetermined number. To choose a keyword combination An initial keyword selection means, a keyword selected by the initial keyword selection means, a display means for classifying and displaying a file by a keyword from an identifier of a text file containing the selected keyword, and a user selects the initial keyword An input means for requesting a classification result different from the classification result of the keyword selected by the selection means;
Classification keyword refinement means for selecting a keyword combination different from the keyword selected by the initial keyword selection means by using a keyword collected by the keyword collection means and a keyword other than the keyword selected by the initial keyword selection means When,
A file set generation means for generating a set selected from the keyword selected by the classification keyword refinement means and an identifier of the text file including the keyword; and a keyword selected by the classification keyword refinement means in the display means And the set of file identifiers generated by the file set generation means are classified and displayed again by keyword , and the initial keyword selection means determines the number of files in which the keyword appears from the keywords collected by the keyword collection means. Arrange them in descending order and select a certain number of keywords from the top to classify them. If the text that includes a given keyword is also included in the lower keywords, assign a file identifier to the lower keywords. This is an information classification device.
[0006]
DETAILED DESCRIPTION OF THE INVENTION
According to an embodiment of the present invention , the present invention relates to a file storage unit that stores a file including text composed of character codes, and a morphological analysis performed on a text portion of the file output by the file storage unit. The result of the morpheme analysis with the input of the morpheme analysis means for outputting together with the identifier of the file, the keyword storage means for storing the keywords for classifying the file, the output of the keyword storage means and the output of the morpheme analysis means A keyword collecting unit that collects a keyword that matches the keyword stored in the keyword storage unit and outputs the keyword together with an identifier of the file; and from a keyword collected by the keyword collecting unit, a file including the keyword Keyword combinations to increase the number And the initial keyword selection means for selecting, and keywords that have been selected by the initial keyword selection means, from the identifier of a text file that contains the keyword that has been selected, and a display means for displaying to classify the files by keyword, user said Using an input means for requesting a classification result different from the classification result of the keyword selected by the initial keyword selection means, and a keyword other than the keyword selected by the initial keyword selection means and the keyword collected by the keyword collection means Classification keyword refinement means for selecting a combination of keywords different from the keyword selected by the initial keyword selection means; the keyword selected by the classification keyword refinement means; and the identifier of the text file containing the keyword as an element Generate a set That the file set generation unit, in the display unit, a keyword selected by the classifying keyword refinement means, said a set of identifiers of the generated files in the file set generation means again displays the classified keywords This is an information classification device.
[0007]
Further, the initial keyword selection means arranges the keywords collected by the keyword collection means in the order of the number of files in which the keywords appear, and selects a certain number of keywords from the top as a keyword to be classified and includes a predetermined keyword. If the text is also included in the lower keyword, a file identifier is assigned to the lower keyword.
[0008]
Further, in the classified keyword refinement means, the keywords collected by the keyword collection means and keywords other than the keyword selected by the initial keyword selection means are arranged in descending order of the number of files in which the keyword appears, and the initial keyword The keywords selected by the selection means are arranged after being arranged in the above order, and a certain number of keywords are selected as refined keywords from the top.
[0011]
FIG. 1 is a block diagram showing the configuration of the entire information classification apparatus according to an embodiment of the present invention . The file storage unit 101 stores a file including text composed of character codes. The morpheme analysis unit 102 performs morpheme analysis on the text portion of the file in the file storage unit 101 and outputs it together with the file identifier. The keyword storage means 103 stores keywords for classifying files. The keyword collection unit 104 collects only the keywords stored in the keyword storage unit 103 from the results of the morphological analysis, and outputs them together with the file identifier. The information classifying means 105 classifies the file identifier by a keyword associated therewith. The display means 106 provides the user with the contents of the file stored in the file storage means 101 according to the classification result. The input means 107 informs the information classification means 105 of the request when the user desires a classification different from the provided classification result.
[0012]
Next, the operation of the present embodiment will be described. As an example, it is assumed that five files describing the ramen restaurant shown in FIG. The identifier of each file is
file1, file2, file3, file4, file5
It is.
[0013]
The morpheme analysis unit 102 performs morpheme analysis on the text portion of the file stored in the file storage unit 101 and outputs it together with the file identifier. FIG. 3 shows a result of extracting only nouns as a result of processing by the morphological analysis unit 102 for the file shown in FIG.
[0014]
The keyword storage means 103 lists keywords used for classification. An example is shown in FIG.
[0015]
The keyword collection unit 104 extracts only the words stored in the keyword storage unit 103 from the output of the morpheme analysis unit 102 and outputs them together with the identifier of the file. FIG. 5 shows the result of the keyword collection unit 104 processing the result of the morphological analysis of FIG.
[0016]
The information classifying unit 105 classifies the set of identifiers of the files having the keywords output by the keyword collecting unit 104 according to the keywords, and outputs them. A block diagram showing a detailed configuration of the information classification means 105 is shown in FIG.
[0017]
The initial keyword selection unit 601 arranges keywords in the order of the number of files in which the keywords appear, and selects a certain number of keywords from the top as a classification keyword set. FIG. 7 shows the result of arranging the outputs of the keyword collecting unit 104 in FIG. 5 from the left in order of the number of appearing files with the keyword as a horizontal axis. If the number of keywords to be selected as a classification keyword set is 2, the set of classification keywords is
{Ramen, soy sauce}
It becomes.
[0018]
The classification keyword refinement unit 602 replaces keywords included in the classification keyword set output by the initial keyword selection unit 601 with other keywords so that they appear in more files. First, the number of files in which keywords included in the classified keyword set appear is used as an evaluation function. Then, the operation of replacing one keyword included in the classified keyword set with a keyword that has not been included in the classified keyword set is repeated as long as the value of the evaluation function increases. In the example of FIG. 7, the classification keyword set is
{Ramen, soy sauce}
Is set to 4, the value of the evaluation function is 4. Replace the keywords “ramen” and “soy sauce taste” included in the classification set with the keywords “baked pork” that have not yet been included in the classification set, and the value of the evaluation function is calculated to be 4. . Therefore, since the value of the evaluation function does not increase, the classified keyword refinement means 602 uses the classified keyword set as
{Ramen, soy sauce}
Output as.
[0019]
The file set generation unit 603 classifies the file identifiers according to the classification keyword set output by the classification keyword refinement unit 602. First, the keywords included in the classification keyword set are arranged in the descending order of the number of files in which the keywords appear, and the set of file identifiers assigned to the keywords is limited to the identifiers of files in which no keyword lower than the keyword appears. . In the example of FIG.
{Ramen, soy sauce}
Then, the file in which the keyword “ramen” appears is
{file1, file3, file5}
However, since the lower keyword “soy sauce taste” appears in file1, the set of file identifiers assigned to each keyword is
Ramen: {file3, file5}
Soy sauce taste: {file1, file4}
It becomes. For a file in which no keyword included in the classified keyword set appears, the special keyword “others” is added to the classified keyword set, and an identifier of the file is assigned to the special keyword “other”. In the example of FIG.
Other: {file2}
As a result of the information classifying unit 105 from the file set generating unit 603,
{Ramen: {file3, file5}, Soy Sauce: {file1, file4}, Others: {file2}}
Is output.
[0020]
The recursive classification control unit 604 is used to further classify the results classified by the file set generation unit 603. That is, by providing the identifier of the already classified file and the set of keywords appearing in the file to the initial keyword selecting means 601, the identifier of the classified file is further classified by the keyword.
[0021]
The display means 106 converts the result of the information classification means 105 into a tree structure, and provides a list of database contents to the user. The output of the information classification means 105 is
{Ramen: {file3, file5}, Soy Sauce: {file1, file4}, Others: {file2}}
In this case, an output result as shown in FIG. 8 is obtained. The user looks at this output result and uses the input means 107 when he wants to request another classification result. The input means 107 is connected to the information classification means 105, and the request is transmitted to the initial keyword selection means 201.
[0022]
The initial keyword selection unit 601 stores a classification keyword set recently selected from the result output by the keyword collection unit 104. When the user's request is transmitted from the input means 107, the keyword column included in the recently selected classification keyword is moved to the lowest-ordered keyword column, and the order is fixed from the highest level. A number of keywords are selected and output as a classification keyword set. In the example of FIG. 7, the classification keyword is {ramen, soy sauce}
Is recently selected, the order is preserved, and the keyword is moved next to the lowest keyword “baked pork” to create a keyword string as shown in FIG. Then select the two keywords from the top, and set the classification keyword set,
{Fried pork, ramen}
Select to output. The processing after the classification keyword refinement device 602 is the same, and as an output of the information classification device,
{Ramen: {file1, file5}, Grilled pork: {file2, file3}, Others: {file4}}
Is output. The display means 106 is provided with a list of database contents as shown in FIG. 10, which is different from the previous classification result of FIG.
[0023]
In the present invention, any type of file can be classified as long as the file includes text composed of character codes. If the file is an HTML file constituting a homepage on the Internet and the file identifier is its URL address, the information classification apparatus of the present invention can be used as a homepage classification system.
[0024]
【The invention's effect】
As is clear from the above description, the present invention uses keywords as concepts and automatically classifies files containing text composed of character codes, so even if a file with a new concept appears, the keywords It is possible to easily update the conceptual system simply by maintaining the database, and to provide a user with a quick list of database contents.
[Brief description of the drawings]
FIG. 1 is a block diagram showing the overall configuration of an information classification apparatus according to an embodiment of the present invention. FIG. 2 is a diagram showing an example of a file storage unit 101 in FIG. 1 for explaining the operation of the embodiment. 3 is a diagram showing an example of the output of the morpheme analyzing means 102 in FIG. 1 for explaining the operation of the embodiment; FIG. 4 is a keyword storing means in FIG. 1 for explaining the operation of the embodiment; FIG. 5 is a diagram showing an example of the output of the keyword collecting means 104 in FIG. 1 for explaining the operation of the embodiment. FIG. 6 is a diagram for explaining the operation of the embodiment. FIG. 7 is a detailed block diagram of the information classification unit 105 in FIG. 1. FIG. 7 is a diagram showing an example of the internal state of the initial keyword selection unit 601 in FIG. 6 for explaining the operation of the embodiment. 1 for explaining the operation of the embodiment FIG. 9 is a diagram showing an example of an internal state of initial keyword selection means 601 in FIG. 6 for explaining the operation of the embodiment. FIG. 10 is a diagram for explaining the operation of the embodiment. The figure which shows an example of the display means 106 of FIG.
101 File storage means 102 Morphological analysis means 103 Keyword storage means 104 Keyword collection means 105 Information classification means 106 Display means 107 Input means 601 Initial keyword selection means 602 Classification keyword refinement means 603 File set generation means 604 Recursive classification control means

Claims (2)

File storage means for storing a file containing text composed of character codes;
Morphological analysis means for performing morphological analysis on the text portion of the file output by the file storage means and outputting together with the identifier of the file;
Keyword storage means for storing keywords for classifying the files;
Using the output of the keyword storage means and the output of the morpheme analysis means as input, collect keywords that match the keywords stored in the keyword storage means from the results of the morpheme analysis and output them together with the identifier of the file Keyword collection means to
Initial keyword selection means for selecting a combination of keywords from the keywords collected by the keyword collection means so that the number of files containing the keywords is greater than a predetermined number ;
Display means for classifying and displaying files by keyword from the keyword selected by the initial keyword selection means and the identifier of the text file containing the selected keyword;
An input means for requesting a classification result different from the classification result of the keyword selected by the initial keyword selection means by the user;
Classification keyword refinement means for selecting a keyword combination different from the keyword selected by the initial keyword selection means by using a keyword collected by the keyword collection means and a keyword other than the keyword selected by the initial keyword selection means When,
File set generation means for generating a set having the keyword selected by the classification keyword refinement means and the identifier of the text file including the keyword as elements;
In the display means, the keyword selected by the classification keyword refinement means and the set of file identifiers generated by the file set generation means are classified and displayed again by keywords,
The initial keyword selection means is
From the keywords collected by the keyword collecting means, arrange the keywords in the order of the number of files in which they appear, select a certain number of keywords from the top to classify them,
An information classification apparatus , wherein when a text including a predetermined keyword is also included in a lower keyword, a file identifier is assigned to the lower keyword . In classification keyword refinement means,
The keywords collected by the keyword collecting means and using keywords other than the keyword selected by the initial keyword selecting means are arranged in descending order of the number of files in which the keyword appears, and the keywords selected by the initial keyword selecting means are arranged. The information classification apparatus according to claim 1, wherein the information is arranged after being arranged in the order, and a certain number of keywords are selected as refined keywords from the top.

Images