JP2013196264A - Similarity search device and computer program and similarity search method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a proper storage place where a file is stored to a user under the consideration of the content of the file.SOLUTION: A file storage part 20 stores a plurality of files. An entry word storage part 25 stores an entry word extracted from each file. A data input part 21 inputs data. An entry word extraction part 24 extracts the entry word from the data input by the data input part 21. A search part 26 searches the entry word matching the entry word extracted by the entry word extraction part 24 from among entry words stored in the entry word storage part 25. A similarity calculation part 27 calculates similarity between each file and the data input by the data input part 21 on the basis of the entry word searched by the search part 26.

Description

  The present invention relates to a similarity search apparatus that searches for a similar file from among a large number of accumulated files.

  In order to store an electronic file in an appropriate file storage area, there is a technique that presents a file storage area with high access frequency and allows the user to select a file storage area.

JP 2010-170472 A JP 2012-33171 A

Since the method for presenting a file storage area with high access frequency does not take into account the contents of the electronic file at all, an appropriate file storage area cannot always be presented.
An object of the present invention is to present a proper storage location for storing a file to a user by considering the contents of the file, for example.

The similarity search device according to the present invention is:
A storage device for storing data, a file storage unit, a headword storage unit, a data input unit, a headword extraction unit, a search unit, and a similarity calculation unit;
The file storage unit stores a plurality of files using the storage device,
The headword storage unit stores headwords extracted from the file for each file stored by the file storage unit using the storage device,
The data input unit inputs data,
The headword extraction unit extracts headwords from the data input by the data input unit,
The search unit searches for a headword that matches the headword extracted by the headword extraction unit from the headwords stored in the headword storage unit,
The similarity calculation unit calculates a similarity between each file stored in the file storage unit and data input by the data input unit based on an entry word searched by the search unit. .

  According to the similarity search device according to the present invention, for example, since the similarity between the data input by the data input unit and the file is calculated, an appropriate storage location for storing the data can be presented.

1 is a diagram illustrating an example of the overall configuration of a file storage system 10 according to Embodiment 1. FIG. FIG. 3 is a diagram illustrating an example of hardware resources of the computer 90 according to the first embodiment. FIG. 3 is a diagram illustrating an example of a functional block configuration of a file storage device 12 according to the first embodiment. FIG. 6 shows an example of file information D20 in the first embodiment. FIG. 6 shows an example of a headword D40 in the first embodiment. FIG. 10 shows an example of index information D50 in the first embodiment. FIG. 6 is a diagram illustrating an example of similarity information D70 in the first embodiment. FIG. 6 is a diagram showing an example of similar output information D80 in the first embodiment. FIG. 3 is a flowchart showing an example of a process flow of the file storage device 12 according to the first embodiment. FIG. 4 is a flowchart showing an example of a flow of a writing process S12 in the first embodiment. FIG. 6 is a flowchart showing an example of a flow of similarity search processing S14 in the first embodiment. FIG. 10 is a diagram illustrating an example of a functional block configuration of a file storage device 12 according to a second embodiment. The figure which shows an example of the headword number information D60 in Embodiment 2. FIG. FIG. 11 is a diagram illustrating an example of similarity information D70 in the second embodiment. FIG. 10 shows an example of similar output information D80 in the second embodiment. FIG. 9 is a flowchart showing an example of a flow of write processing S12 in the second embodiment. FIG. 11 is a flowchart showing an example of a flow of similarity search processing S14 in the second embodiment. FIG. 10 is a diagram illustrating an example of a functional block configuration of a file storage device 12 according to a third embodiment. FIG. 10 shows an example of fitness information D90 in the third embodiment. FIG. 14 is a flowchart showing an example of a process flow of the file storage device 12 according to the third embodiment. FIG. 10 is a flowchart showing an example of a flow of location search processing S15 in the third embodiment.

Embodiment 1 FIG.
The first embodiment will be described with reference to FIGS.

  FIG. 1 is a diagram showing an example of the overall configuration of a file storage system 10 in this embodiment.

The file storage system 10 includes, for example, a file editing device 11 and a file storage device 12.
The file storage device 12 stores a large number of files. A file is a unit in which electronic data stored in the file storage device 12 is managed by an operating system (OS) or the like.
The file editing device 11 is a device for browsing and editing a file stored in the file storage device 12 in accordance with a user operation. Further, the file editing device 11 inputs new data in accordance with the user's operation, and stores the input data in the file storage device 12 as a new file.

  FIG. 2 is a diagram illustrating an example of hardware resources of the computer 90 in this embodiment.

The file editing device 11 and the file storage device 12 are, for example, a computer 90.
The computer 90 includes, for example, a processing device 91, an input device 92, an output device 93, and a storage device 94.

The storage device 94 stores a computer program executed by the processing device 91, data processed by the processing device 91, and the like. The storage device 94 is, for example, an internal storage device such as a semiconductor memory, or an external storage device such as a magnetic disk device or an optical disk device.
The processing device 91 processes the data by executing the computer program stored in the storage device 94 and controls the entire computer 90.
The input device 92 inputs information from outside and converts it into data to be processed by the processing device 91. The data converted by the input device 92 may be processed directly by the processing device 91 or may be stored temporarily by the storage device 94. The input device 92 is, for example, an operation input device such as a keyboard or a mouse, a voice input device such as a microphone, an image input device such as a camera or a scanner, a sensor, an analog-digital conversion device, or a reception device.
The output device 93 converts the data processed by the processing device 91 and the data stored in the storage device 94 and outputs them to the outside. The output device 93 is, for example, a sound output device such as a speaker, an image display device, a printing device, a digital / analog conversion device, a transmission device, or the like.

Functional blocks such as the file storage device 12 are realized by the processing device 91 executing a computer program, for example. However, these functional blocks may be realized by other electrical configurations or mechanical configurations.
Further, the file storage device 12 or the like may be configured by a plurality of computers 90 instead of a single computer 90. Conversely, a single computer 90 may constitute both the file editing device 11 and the file storage device 12.

  FIG. 3 is a diagram showing an example of the functional block configuration of the file storage device 12 in this embodiment.

  The file storage device 12 includes, for example, a file storage unit 20, an instruction input unit 22, a data input unit 21, a file output unit 23, a headword extraction unit 24, a headword storage unit 25, and a search unit 26. A similarity calculation unit 27 and a similarity output unit 28.

  The file storage unit 20 stores a plurality of files using the storage device 94. The file stored in the file storage unit 20 is identified by a file identifier such as a document ID, for example. In addition, the file name stored by the file storage unit 20 is given a file name named by the user or the like so that the user can easily understand the contents of the file. In addition, the file stored in the file storage unit 20 is stored in a storage location such as a folder designated by the user so that the user can sort and organize the file. The storage location may reflect a physical difference between the storage devices 94 or may be a logical one that is unrelated to the physical difference between the storage devices 94.

  The data input unit 21 inputs data from the file editing device 11.

The instruction input unit 22 inputs an instruction by the user from the file editing apparatus 11 using the input device 92. Examples of instructions by the user include a write instruction, a read instruction, and a similarity search instruction.
The write instruction is an instruction to store the data input by the data input unit 21 in the file storage unit 20 as a file. The write instruction includes, for example, the file name and storage location of the file stored in the file storage unit 20.
The read instruction is an instruction to output the contents of the file stored in the file storage unit 20. The read instruction includes, for example, the file name and storage location of the output file.
The similarity search instruction is an instruction to search for a file having contents similar to the data input by the data input unit 21 from the files stored in the file storage unit 20.

  The file output unit 23 uses the output device 93 to output the contents of the file stored in the file storage unit 20.

The headword extraction unit 24 uses the processing device 91 to extract headwords from the data input by the data input unit 21 and the contents of the file stored by the file storage unit 20. A headword is a keyword representing the contents of the data or file. Examples of the headword extraction method include a word division method and an N-gram method.
The word division method is a method in which data is divided into words and used as a headword. For example, the headword extraction unit 24 stores a word dictionary in advance, and extracts a word as a headword when a word registered in the word dictionary appears in the data.
The N-gram method is a method of cutting out a character string of a predetermined length regardless of the meaning. For example, the headword extraction unit 24 extracts data from the first character to the Nth character as the first headword, and extracts from the second character to the (N + 1) th character as the second headword. Thus, the headword is extracted while shifting character by character.
The method by which the headword extraction unit 24 extracts headwords is not limited to the word division method or the N-gram method, and may be another method.

  The headword storage unit 25 stores, for each file, the headword extracted by the headword extraction unit 24 from the file stored by the file storage unit 20 using the storage device 94.

  The search unit 26 uses the processing device 91 to search for a word that matches the headword extracted by the headword extraction unit 24 from the data input by the data input unit 21 from the headwords stored in the headword storage unit 25. Search for.

The similarity calculation unit 27 uses the processing device 91 to determine between the data input by the data input unit 21 and each file stored in the file storage unit 20 based on the headword searched by the search unit 26. Calculate similarity.
For example, for each file, the similarity calculation unit 27 counts the number of headwords that match the headword extracted from the data input by the data input unit 21 among the headwords extracted from the file. And In this case, the greater the similarity value, the more similar the data input by the data input unit 21 and the content of the file.

The similarity output unit 28 is similar to the data input by the data input unit 21 from among the files stored in the file storage unit 20 based on the similarity calculated by the similarity calculation unit 27 using the processing device 91. Files that are The similar output unit 28 uses the output device 93 to output information such as the file name and storage location of the extracted file.
For example, the similarity output unit 28 extracts a predetermined number of files in order from the highest similarity calculated by the similarity calculation unit 27.

  FIG. 4 is a diagram showing an example of the file information D20 in this embodiment.

  The file storage unit 20 stores, for example, file information D20. The file information D20 includes, for example, a plurality of file identifiers D21, a plurality of file names D22, a plurality of storage locations D23, and a plurality of contents data D24. One file identifier D21 is associated with one file name D22, one storage location D23, and one content data D24. The file identifier D21 is an identifier that uniquely identifies a file. The file name D22 represents the name of the file. The storage location D23 represents the storage location of the file. Content data D24 represents the content of the file.

  FIG. 5 is a diagram showing an example of the headword D40 in this embodiment.

  In this example, the headword extraction unit 24 extracts headwords by the N-gram method (N = 2). The headword extraction unit 24, for example, from a file whose content data D24 is “Today's sunny weather. Ten headwords D40, “Ming”, “Tomorrow”, “Sunlight” and “rainy weather” are extracted. “Ten”, “Nari”, and “Ri.” Appear twice in the content data D24. In this example, the headword extraction unit 24 does not extract the same headword twice, but different headlines. Extract words only. The headword extraction unit 24 may be configured to extract the same headword a plurality of times.

  FIG. 6 is a diagram showing an example of the index information D50 in this embodiment.

The headword storage unit 25 stores, for example, index information D50. The index information D50 includes, for example, a plurality of headwords D51 and a plurality of file identifiers D52. One file identifier D52 is associated with one headword D51. The headword D51 is a headword extracted from the file by the headword extraction unit 24. The file identifier D52 is the file identifier of the file from which the headword extraction unit 24 has extracted the headword D51.
The headword extraction unit 24 normally extracts a plurality of headwords from one file. For this reason, the file identifier D52 includes duplicates. In addition, the headword extraction unit 24 may extract the same headword from different files. For this reason, the headword D51 also includes overlapping words.

Note that when the headword extraction unit 24 is configured to extract a plurality of the same headwords from one file, the index information D50 includes, for example, the headword extraction unit 24 from the file identified by the file identifier D52. The structure including the extraction number showing the number which extracted D51 may be sufficient.
Further, the index information D50 is not limited to the configuration shown in this example, and any configuration may be used as long as the file from which the headword is extracted can be searched from the headword. For example, the index information D50 may be configured so that overlapping entries of the entry word D51 are combined into one, and a list of file identifiers of the file from which the entry word is extracted is included instead of the file identifier D52.

  FIG. 7 is a diagram showing an example of the similarity information D70 in this embodiment.

For example, the similarity calculation unit 27 generates similarity information D70. The similarity information D70 includes, for example, a plurality of file identifiers D71 and a plurality of hit counts D72. One hit identifier D72 is associated with one file identifier D71. The file identifier D71 is a file identifier of a file from which the same headword as the data input by the data input unit 21 is extracted. The hit count D72 represents the number of matching headwords between the headword extracted from the data input by the data input unit 21 and the headword extracted from the file identified by the file identifier D71.
Even when the same headword appears multiple times in the data input by the data input unit 21 or when the same headword appears multiple times in one file, the similarity calculation unit 27 determines the number of hits as one. Count as.

  When the headword extraction unit 24 is configured to extract a plurality of the same headwords from one file, a headword is extracted a times from the data input by the data input unit 21 and identified by the file identifier D71. Assuming that the file is extracted b times from the file, the similarity calculation unit 27 may be configured to count the number of hits a × b (or b) with only one headword. As a result, when headwords with a large number of appearances are common, the similarity calculated by the similarity calculation unit 27 increases.

  FIG. 8 is a diagram showing an example of the similar output information D80 in this embodiment.

  The similar output unit 28 outputs, for example, similar output information D80. The similar output information D80 includes, for example, a plurality of file identifiers D81, a plurality of file names D82, a plurality of storage locations D83, and a plurality of hit counts D84. One file identifier D81 is associated with one file name D82, one storage location D83, and one hit count D84. The file identifier D81 is a file identifier of a file similar to the data input by the data input unit 21. The file name D82 represents the file name of the file. A storage location D83 represents a storage location of the file. The hit count D84 represents the number of matching headwords between the headword extracted from the data input by the data input unit 21 and the headword extracted from the file identified by the file identifier D71.

  FIG. 9 is a flowchart showing an example of the processing flow of the file storage device 12 in this embodiment.

  The file storage device 12 executes, for example, an instruction input process S11, a write process S12, a read process S13, and a similarity search process S14.

In the instruction input step S <b> 11, the instruction input unit 22 inputs an instruction from the user from the file editing device 11.
If the input instruction is a write instruction, the instruction input unit 22 proceeds to the write process S12.
If the input instruction is a read instruction, the instruction input unit 22 advances the process to the read process S13.
If the input instruction is a similarity search instruction, the instruction input unit 22 advances the process to the similarity search process S14.

  In the writing process S <b> 12, the file storage device 12 stores the data input by the data input unit 21 in the file storage unit 20. Details of the writing process S12 will be described later. After the end of the writing process S12, the instruction input unit 22 returns the process to the instruction input process S11 and waits for the next instruction.

In the read process S13, the file storage unit 20 acquires the file name and the storage location from the read instruction input by the instruction input unit 22 in the instruction input step S11. From the stored file information D20, the file storage unit 20 is associated with the file name D22 that matches the acquired file name, and with the file identifier D21 that is associated with the storage location D23 that matches the acquired storage location. To extract.
When there is a file identifier D21 that satisfies the condition, the file storage unit 20 acquires content data D24 associated with the extracted file identifier D21. The file output unit 23 outputs the content data D24 acquired by the file storage unit 20 to the file editing apparatus 11.
If the file identifier D21 that satisfies the condition does not exist, the file output unit 23 outputs an error to the file editing apparatus 11.
The instruction input unit 22 returns the process to the instruction input step S11 and waits for the next instruction.

  In the similarity search process S <b> 14, the file storage unit 20 searches the file stored in the file storage unit 20 for a file similar to the data input by the data input unit 21. Details of the similarity search processing S14 will be described later. After the similarity search process S14 ends, the instruction input unit 22 returns the process to the instruction input step S11 and waits for the next instruction.

  FIG. 10 is a flowchart showing an example of the flow of the writing process S12 in this embodiment.

  The writing process S12 includes, for example, an overwrite determination step S21, a headword deletion step S22, a data input step S23, a file storage step S24, a headword extraction step S25, and a headword storage step S26.

In the overwrite determination step S21, the file storage unit 20 acquires the file name and the storage location from the write instruction input by the instruction input unit 22 in the instruction input process S11. The file storage unit 20 corresponds to the storage location D23 that is associated with the file name D22 that matches the file name acquired from the write instruction, and that matches the storage location acquired from the write instruction, from the stored file information D20. The attached file identifier D21 is extracted.
If the file identifier D21 that satisfies the condition exists, the file is overwritten. The file storage unit 20 deletes the content data D24 associated with the extracted file identifier D21. The file storage unit 20 proceeds to the entry word deletion step S22.
If the file identifier D21 that satisfies the condition does not exist, a new file is created. The file storage unit 20 generates a new file identifier and stores it as a file identifier D21. The file storage unit 20 stores the file name acquired from the write instruction as the file name D22 associated with the file identifier D21. The file storage unit 20 stores the storage location acquired from the write instruction as the storage location D23 associated with the file identifier D21. The file storage unit 20 proceeds to the data input process S23.

In the headword deletion step S22, the headword storage unit 25 extracts a file identifier D52 that matches the file identifier D21 extracted by the file storage unit 20 in the overwrite determination step S21 from the stored index information D50.
When there is a file identifier D52 that satisfies the condition, the headword storage unit 25 deletes all of the extracted file identifier D52 and the headword D51 associated therewith.

In the data input step S <b> 23, the data input unit 21 inputs data from the file editing device 11. The data input unit 21 inputs data one by one in order.
When the end of the data is reached and there are no more characters to be input, the data input unit 21 ends the writing process S12.
If the end of the data has not yet been reached and data for one character has been input, the data input unit 21 proceeds to the file storage step S24.

  In the file storage step S24, the file storage unit 20 adds the data for one character input in the data input step S23 to the end of the content data D24 associated with the file identifier D21 extracted or generated in the overwrite determination step S21. Remember.

  In the headword extraction step S25, the headword extraction unit 24 traces the characters input by the data input unit 21 in the data input step S23 N times and sets them as headwords composed of N characters. However, N is an integer of 1 or more.

In the headword storage step S26, the headword storage unit 25 uses the headword D51 that matches the headword extracted by the headword extraction unit 24 in the headword extraction step S25 from the stored index information D50, and the overwrite determination step. In S21, a set in which the file identifier D52 that matches the file identifier D21 extracted or generated by the file storage unit 20 is associated is extracted.
When there is a set of a headword D51 and a file identifier D52 that satisfy the condition, the headword has already been extracted.
When there is no combination of the headword D51 and the file identifier D52 that satisfy the condition, the headword is not extracted. The headword storage unit 25 stores the headword extracted by the headword extraction unit 24 in the headword extraction step S25 as a headword D51. The headword storage unit 25 stores the file identifier D21 extracted or generated by the file storage unit 20 in the overwrite determination step S21 as the file identifier D52 associated with the headword D51.
The data input unit 21 returns the process to the data input step S23 and inputs the next character.

  As described above, when storing a new file, the file storage device 12 extracts an entry word from the file and creates an index in advance.

  Instead of creating an index while inputting data, the index may be created after first inputting and storing the data.

  FIG. 11 is a flowchart showing an example of the flow of the similarity search process S14 in this embodiment.

  The similarity search processing S14 includes, for example, an initialization step S40, a data input step S41, a headword extraction step S42, a headword search step S43, a file selection step S44, a coincidence counting step S45, and a rearrangement step S50. And a file selection step S51 and a similar output step S52.

In the initialization step S40, the headword extraction unit 24 initializes a list of extracted headwords. For example, the headword extraction unit 24 stores an empty list as a list of extracted headwords.
The similarity calculation unit 27 initializes the similarity information D70. For example, the similarity calculation unit 27 deletes the stored similarity information D70.

In the data input step S <b> 41, the data input unit 21 inputs data from the file editing device 11. The data input unit 21 inputs data one by one in order.
When the end of the data is reached and there are no more characters to be input, the data input unit 21 proceeds to the rearrangement step S50.
If the end of the data has not been reached yet and one character's worth of data has been input, the data input unit 21 proceeds to the headword extraction step S42.

In the headword extraction step S42, the headword extraction unit 24 goes back N times for the character input in the data input step S41 and sets it as a headword composed of N characters. However, N is an integer of 1 or more.
The headword extraction unit 24 determines whether or not the current headword is present in the extracted list of headwords.
If the current headword is present in the extracted headword list, the instruction input unit 22 returns the process to the data input step S41 and inputs the next character.
If the current headword does not exist in the list of extracted headwords, the headword extraction unit 24 stores the current headword in addition to the extracted headword list. The headword extraction unit 24 proceeds to the headword search step S43.

In the headword search step S43, the search unit 26 selects a headword D51 that matches the headword extracted by the headword extraction unit 24 in the headword extraction step S42 from the index information D50 stored in the headword storage unit 25. Extract.
If there is a headword D51 that satisfies the condition, the search unit 26 advances the process to the coincidence counting step S45.
If there is no headword D51 that satisfies the condition, the data input unit 21 returns the process to the data input step S41 and inputs the next character.

In the file selection step S44, the similarity calculation unit 27 sets a file identifier D52 that has not yet been selected from the file identifiers D52 associated with the headword D51 extracted by the search unit 26 in the headword search step S43. Select one.
If all the file identifiers D52 associated with the headword D51 extracted by the search unit 26 have been selected and there is no file identifier D52 that has not yet been selected, the data input unit 21 returns the process to the data input step S41. Enter the next character.
When there is a file identifier D52 that has not yet been selected among the file identifiers D52 that are associated with the headword D51 extracted by the search unit 26, the similarity calculation unit 27 selects the file identifier D52 that has not yet been selected. Then, one file identifier D52 is selected.

In the coincidence counting step S45, the similarity calculating unit 27 extracts a file identifier D71 that matches the file identifier D52 selected in the file selection step S44 from the stored similarity information D70.
When there is a file identifier D71 that matches the file identifier D52, the similarity calculation unit 27 adds 1 to the number of hits D72 associated with the file identifier D71.
When there is no file identifier D71 that matches the file identifier D52, the similarity calculation unit 27 stores the file identifier D52 selected in the file selection step S44 as the file identifier D71. The similarity calculation unit 27 stores 1 as the hit count D72 associated with the file identifier D71.
The similarity calculation unit 27 returns the process to the file selection step S44 and selects the next file identifier D52.

  In the rearrangement step S50, the similarity output unit 28 rearranges the similarity information D70 stored by the similarity calculation unit 27 in descending order of the hit count D72.

In the file selection step S51, the similarity output unit 28 selects a file identifier D71 that has not been selected from the similarity information D70 rearranged in the rearrangement step S50.
When all the file identifiers D71 have been selected and there is no file identifier D71 that has not yet been selected, or when the number of selected file identifiers D71 reaches a predetermined number, the similar output unit 28 performs similar search processing. S14 ends.
If there is a file identifier D71 that has not yet been selected, and the number of selected file identifiers D71 has not yet reached a predetermined number, the similar output unit 28 associates the file identifier D71 that has not yet been selected. The file identifier D71 having the largest hit count D72 is selected.

In the similar output step S52, the similar output unit 28 selects the file name D22 associated with the file identifier D21 that matches the file identifier D71 selected in the file selection step S51 from the file information D20 stored in the file storage unit 20. And the storage location D23 is acquired. The similar output unit 28 outputs the file identifier D71 selected in the file selection step S51 as the file identifier D81. The similar output unit 28 outputs the file name D22 acquired from the file information D20 as the file name D82 associated with the file identifier D81. The similar output unit 28 outputs the storage location D23 acquired from the file information D20 as the storage location D83 associated with the file identifier D81. The similar output unit 28 outputs the hit count D72 associated with the file identifier D71 selected in the file selection step S51 as the hit count D84 associated with the file identifier D81.
The similar output unit 28 returns the process to the file selection step S51 and selects the next file identifier D71.

  As described above, a file having a large number of matching headwords is determined as a similar file. Since an index is created in advance, a similar file can be quickly found even if the number of files stored in the file storage unit 20 is large.

  Instead of searching while inputting data, a configuration may be used in which searching is performed after data is first input and temporarily stored.

  As described above, the headword is extracted when the file is stored. The headword of the file to be stored is added to the existing index (index information D50), and the number of bits is totaled by document ID (file identifier). Information such as the file name and storage location is added from the existing file information D20 and output. Thereby, a similar file can be acquired, and can be referred to and used.

Embodiment 2. FIG.
The second embodiment will be described with reference to FIGS.
In addition, about the part which is common in Embodiment 1, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  In this embodiment, another example of the similarity calculated by the similarity calculation unit 27 will be described.

  FIG. 12 is a diagram showing an example of the functional block configuration of the file storage device 12 in this embodiment.

  In addition to the headword extracted by the headword extraction unit 24 using the storage device 94, the headword storage unit 25 further calculates the number of headwords extracted from each file by the headword extraction unit 24 for each file. To remember.

  The similarity calculation unit 27 uses the processing device 91 to determine the ratio of headwords that match the headwords extracted from the data input by the data input unit 21 among the headwords extracted from the files for each file. Let the similarity be. For example, the similarity calculation unit 27 calculates the similarity described in the first embodiment, and calculates a quotient obtained by dividing the calculated similarity by the number of headwords stored for the file by the headword storage unit 25. Thus, the similarity is used in this embodiment.

  FIG. 13 is a diagram showing an example of headword number information D60 in this embodiment.

  The headword storage unit 25 stores headword number information D60, for example. The headword number information D60 includes, for example, a plurality of file identifiers D61 and a plurality of total headword numbers D62. One file identifier D61 is associated with one total headword number D62. The file identifier D61 is a file identifier of the file stored in the file storage unit 20. The total number of headwords D62 represents the total number of headwords extracted by the headword extraction unit 24 from the file.

  FIG. 14 is a diagram showing an example of the similarity information D70 in this embodiment.

  The similarity information D70 further includes a plurality of hit ratios D73 in addition to the data described in the first embodiment. One hit ratio D73 is associated with one file identifier D71. The hit ratio D73 is the headword extracted by the headword extraction unit 24 from the data input by the data input unit 21 among the headwords extracted by the headword extraction unit 24 from the file identified by the associated file identifier D71. Represents the percentage of headwords that match.

  FIG. 15 is a diagram showing an example of similar output information D80 in this embodiment.

  The similar output information D80 further includes a plurality of hit ratios D85 in addition to the data described in the first embodiment. One hit ratio D85 is associated with one file identifier D81. The hit ratio D85 is the headword extracted by the headword extraction unit 24 from the data input by the data input unit 21 among the headwords extracted by the headword extraction unit 24 from the file identified by the associated file identifier D71. Represents the percentage of headwords that match.

  FIG. 16 is a flowchart showing an example of the flow of the writing process S12 in this embodiment.

  In addition to the steps described in the first embodiment, the writing process S12 further includes a headword counting step S27 and a headword number storage step S28.

If the headword has been extracted in the headword storage step S26, the data input unit 21 returns to the data input step S23 and inputs the next character.
When the headword has not been extracted, the headword storage unit 25 advances the processing to the headword counting step S27.

In the headword counting step S27, the headword storage unit 25 adds 1 to the total number of headwords.
The data input unit 21 returns to the data input step S23 and inputs the next character.

  In the data input step S23, when the end of the data is reached, the data input unit 21 advances the process to the headword number storage step S28.

  In the headword number storage step S28, the headword storage unit 25 stores the file identifier D21 extracted or generated in the overwrite determination step S21 as the file identifier D61. The headword storage unit 25 stores the calculated total headword count as the total headword count D62 associated with the file identifier D61.

  Thus, the total number of headwords is calculated in advance when creating an index.

  FIG. 17 is a flowchart showing an example of the flow of the similarity search process S14 in this embodiment.

  The similarity search process S14 further includes a file selection process S46 and a ratio calculation process S47 in addition to the processes described in the first embodiment.

  In the data input step S41, when the end of the data is reached, the data input unit 21 advances the process to the file selection step S46.

In the file selection step S46, the similarity calculation unit 27 selects one file identifier D71 that has not been selected from the stored similarity information D70.
When all the file identifiers D71 have been selected and there is no file identifier D71 that has not yet been selected, the similarity calculation unit 27 advances the processing to the rearrangement step S50.
When there is a file identifier D71 that has not been selected, the similarity calculation unit 27 selects one file identifier D71 from among the file identifiers D71 that have not yet been selected, and proceeds to the ratio calculation step S47.

In the ratio calculation step S47, the similarity calculation unit 27 extracts the file identifier D61 that matches the file identifier D71 selected in the file selection step S46 from the headword number information D60 stored in the headword storage unit 25. The similarity calculation unit 27 acquires the total number of headwords D62 associated with the file identifier D61 extracted from the headword number information D60. The similarity calculation unit 27 calculates a quotient obtained by dividing the hit number D72 associated with the file identifier D71 selected in the file selection step S46 by the total headword number D62 acquired from the headword number information D60. The similarity calculation unit 27 stores the calculated quotient as the hit ratio D73 associated with the file identifier D71 selected in the file selection step S46.
The similarity calculation unit 27 returns the process to the file selection step S46, and selects the next file identifier D71.

In this way, the number of hits is not directly used as the similarity, but the quotient obtained by dividing the number of hits by the total number of headwords is used as the similarity. As the number of headwords extracted by the headword extraction unit 24 from one file increases, the number of hits tends to increase. By using the hit ratio as the similarity, a similar file can be found without depending on the number of headwords extracted by the headword extraction unit 24 from one file.
In addition, since the total number of headwords is calculated in advance, a similar file can be quickly found even when the number of files stored in the file storage unit 20 is large.

Embodiment 3 FIG.
The third embodiment will be described with reference to FIGS.
Note that portions common to Embodiment 1 or Embodiment 2 are denoted by the same reference numerals and description thereof is omitted.

  In this embodiment, a configuration for extracting a storage location candidate for storing data as a file based on the similarity will be described.

  FIG. 18 is a diagram showing an example of the functional block configuration of the file storage device 12 in this embodiment.

  In addition to the configuration described in the second embodiment, the file storage device 12 further includes a fitness calculation unit 29 and a storage location candidate extraction unit 30.

The fitness level calculation unit 29 uses the processing device 91 and based on the similarity level calculated by the similarity level calculation unit 27, the fitness level representing a level suitable as a storage location for storing data input by the data input unit 21. Is calculated for each storage location where the file storage unit 20 stores the file.
For example, for each storage location, the suitability calculation unit 29 sums up the similarities calculated by the similarity calculation unit 27 for all the files stored in the storage location to obtain the suitability of the storage location. .

The storage location candidate extraction unit 30 uses the processing device 91 to select the data input unit 21 from the storage locations in which the file storage unit 20 stores files based on the fitness calculated by the fitness calculation unit 29. The candidate of the storage place which memorize | stores the data input by is extracted. The storage location candidate extraction unit 30 uses the output device 93 to output the extracted storage location candidates.
For example, the storage location candidate extraction unit 30 extracts a predetermined number of storage locations as storage location candidates in descending order of the fitness calculated by the fitness calculation unit 29.

  FIG. 19 is a diagram showing an example of the fitness information D90 in this embodiment.

  The fitness level calculation unit 29 generates, for example, fitness level information D90. The fitness information D90 includes, for example, a plurality of storage locations D91 and a plurality of fitness levels D92. One fitness level D92 is associated with one storage location D91. The storage location D91 represents a storage location where the file storage unit 20 stores a file. The fitness level D92 represents the fitness level calculated by the fitness level calculating unit 29 for the associated storage location D91.

  FIG. 20 is a flowchart showing an example of the processing flow of the file storage device 12 in this embodiment.

  In addition to the processing described in the first embodiment, the file storage device 12 further executes a location search process S15.

In the instruction input step S11, when the instruction input by the instruction input unit 22 is a location search instruction, the instruction input unit 22 advances the process to the location search process S15.
The location search instruction is an instruction to search for a candidate for a storage location where the data input by the data input unit 21 should be stored.

  FIG. 21 is a flowchart showing an example of the flow of the location search processing S15 in this embodiment.

  The location search process S15 includes, for example, an initialization step S40, a data input step S41, a headword extraction step S42, a headword search step S43, a file selection step S44, a coincidence counting step S45, and a file selection step S46. And a ratio calculation step S47, a tabulation step S48, a rearrangement step S50, a location selection step S53, and a location output step S54. Of these steps, the steps denoted by the same reference numerals as those of the similar search processing S14 described in the second embodiment are the same as the steps of the similar search processing S14.

  After the ratio calculation step S47 is completed, the similarity calculation unit 27 advances the processing to the counting step S48.

In the counting step S48, the fitness level calculation unit 29 extracts the file identifier D21 that matches the file identifier D71 selected by the similarity level calculation unit 27 in the file selection step S46 from the file information D20 stored in the file storage unit 20. To do. The fitness level calculation unit 29 acquires the storage location D23 associated with the extracted file identifier D21.
The fitness level calculation unit 29 extracts a storage location D91 that matches the acquired storage location D23 from the stored fitness level information D90.
When there is a storage location D91 that satisfies the condition, the fitness calculation unit 29 adds the similarity calculated by the similarity calculation unit 27 in the ratio calculation step S47 to the fitness D92 associated with the extracted storage location D91. .
When there is no storage location D91 that satisfies the condition, the fitness level calculation unit 29 stores the storage location D23 acquired from the file storage unit 20 as the storage location D91. The fitness level calculation unit 29 stores the similarity level calculated by the similarity level calculation unit 27 in the ratio calculation step S47 as the fitness level D92 associated with the storage location D91.
The similarity calculation unit 27 returns the process to the file selection step S46, and selects the next file identifier D71.

  In the rearrangement step S50, the storage location candidate extraction unit 30 rearranges the fitness level information D90 stored by the fitness level calculation unit 29 in descending order of the fitness level D92.

In the location selection step S53, the storage location candidate extraction unit 30 selects a storage location D91 that has not been selected from the fitness information D90 rearranged in the rearrangement step S50.
When all the storage locations D91 have been selected and there is no storage location D91 that has not yet been selected, or when the number of selected storage locations D91 reaches a predetermined number, the storage location candidate extraction unit 30 The search process S15 is terminated.
If there is a storage location D91 that has not yet been selected, and the number of selected storage locations D91 has not yet reached the predetermined number, the storage location candidate extraction unit 30 selects the storage location D91 that has not yet been selected, A storage location D91 having the highest matching degree D92 is selected.

In the location output step S54, the storage location candidate extraction unit 30 outputs the storage location D91 selected in the location selection step S53 and the fitness D92 associated with the storage location D91.
The storage location candidate extraction unit 30 returns the processing to the location selection step S53 and selects the next storage location D91.

  As described above, the suitability of the storage location is calculated based on the similarity calculated for the file stored in a certain storage location. Therefore, the storage location that contains many similar files is the storage location that stores the data. Extracted as a candidate. Since a storage location containing many similar files is extracted as a storage location candidate for storing data, a storage location candidate suitable for the contents of the data can be presented.

  The degree-of-fit calculation unit 29 averages the similarities calculated by the degree-of-similarity calculation unit 27 for all the files stored in the respective storage locations, and obtains the suitability of the storage locations. It may be a configuration.

  Further, the similarity calculation unit 27 may be configured so that the number of hits is not the hit ratio but the similarity.

  In many cases, files that the user has determined to be closely related files are placed in the same storage location. For this reason, even if the number of matching headwords is small, if many files with high similarity are stored in the same storage location, it is highly likely that the file is also a file having some relationship. Therefore, the storage location output by the storage location candidate extraction unit 30 in response to the location search instruction can be used not only for the user to determine the storage location for storing the data, but also when searching for a closely related file. .

Further, in response to the similarity search instruction, the similarity output unit 28 stores a file having a low similarity calculated by the similarity calculation unit 27 in a storage location having a high fitness calculated by the fitness calculation unit 29. The file may be output as a file similar to the data input by the data input unit 21. For example, the similarity output unit 28 calculates the sum of the product obtained by multiplying the fitness calculated by the fitness calculation unit 29 by a predetermined coefficient (for example, 0.1) and the similarity calculated by the similarity calculation unit 27. Then, files are output in descending order of the calculated sum.
This makes it possible to find files that are close to each other even if the degree of similarity is low.

  As described above, the configuration described in each embodiment is an example, and another configuration may be used. For example, the structure which combined the structure demonstrated in different embodiment may be sufficient, and the structure which replaced the structure of the non-essential part with the other structure may be sufficient.

The similarity search device (file storage device 12) described above includes a storage device (94) for storing data, a file storage unit (20), a headword storage unit (25), a data input unit (21), The headword extraction unit (24), the search unit (26), and the similarity calculation unit (27) are included.
The file storage unit stores a plurality of files using the storage device.
The headword storage unit stores headwords extracted from the file for each file stored in the file storage unit using the storage device.
The data input unit inputs data.
The headword extraction unit extracts headwords from the data input by the data input unit.
The search unit searches for a headword that matches the headword extracted by the headword extraction unit from the headwords stored in the index storage unit.
The similarity calculation unit calculates the similarity between each file stored in the file storage unit and the data input by the data input unit, based on the headword extracted by the search unit.

  Since the similarity between the data input by the data input unit and the file is calculated, an appropriate storage location for storing the data can be presented.

The similarity search device (12) further includes a fitness calculation unit (29) and a storage location candidate extraction unit (30).
The file storage unit (20) stores each of the plurality of files in any of a plurality of storage locations.
For each of the plurality of storage locations, the fitness calculation unit is configured so that the data input unit determines whether the file storage unit stores the file stored in the storage location based on the similarity calculated by the similarity calculation unit (27). Calculate the suitability of the storage location for the entered data.
The storage location candidate extraction unit selects a storage location candidate of the data input by the data input unit (21) from the plurality of storage locations based on the fitness calculated by the storage location fitness calculation unit. Extract.

  Since the fitness is calculated based on the similarity, an appropriate storage location for storing data can be presented.

  The fitness level calculation unit (29) calculates the total or average of the similarities calculated by the similarity level calculation unit (27) for the files stored in the storage location to obtain the fitness level of the storage location.

  Since the relevance is the sum or average of the similarities, an appropriate storage location for storing data can be presented.

  The similarity calculation unit (27) is configured so that the headword extraction unit (24) uses the data input by the data input unit (21) among the headwords extracted from the file stored in the file storage unit (20). The number or ratio of headwords that match the extracted headwords is calculated and used as the similarity between the file and the data.

  Since the number or ratio of matching headwords is the similarity, an appropriate storage location for storing data can be presented.

  10 file storage system, 11 file editing device, 12 file storage device, 20 file storage unit, 21 data input unit, 22 instruction input unit, 23 file output unit, 24 headword extraction unit, 25 headword storage unit, 26 search unit , 27 Similarity calculation unit, 28 Similar output unit, 29 Conformity calculation unit, 30 Storage location candidate extraction unit, 90 Computer, 91 Processing device, 92 Input device, 93 Output device, 94 Storage device, D20 File information, D21, D52, D61, D71, D81 File identifier, D22, D82 File name, D23, D83, D91 Storage location, D24 Content data, D40, D51 Headword, D50 Index information, D60 Headword number information, D62 Total headword number, D70 Similar Degree information, D72, D84 hits , D73, D85 hit ratio, D80 similarity output information, D90 fitness information, D92 fitness, S11 instruction input process, S12 write process, S13 read process, S14 location search process, S15 location search process, S21 overwrite determination process, S22 Headword deletion step, S23, S41 Data input step, S24 File storage step, S25, S42 Headword extraction step, S26 Headword storage step, S27 Headword count step, S28 Headword count storage step, S40 Initialization step, S43 Heading Word search step, S44, S46, S51 File selection step, S45 coincidence counting step, S47 ratio calculation step, S48 tabulation step, S50 rearrangement step, S52 similar output step, S53 place selection step, S54 place output step.

Claims (7)

A storage device for storing data, a file storage unit, a headword storage unit, a data input unit, a headword extraction unit, a search unit, and a similarity calculation unit;
The file storage unit stores a plurality of files using the storage device,
The headword storage unit stores headwords extracted from the file for each file stored by the file storage unit using the storage device,
The data input unit inputs data,
The headword extraction unit extracts headwords from the data input by the data input unit,
The search unit searches for a headword that matches the headword extracted by the headword extraction unit from the headwords stored in the headword storage unit,
The similarity calculation unit calculates a similarity between each file stored in the file storage unit and data input by the data input unit based on an entry word searched by the search unit. Similarity search device. The similarity search device further includes a fitness calculation unit and a storage location candidate extraction unit,
The file storage unit stores each of the plurality of files in any of a plurality of storage locations,
For each of the plurality of storage locations, the fitness calculation unit is configured to receive data input by the data input unit based on the similarity calculated by the similarity calculation unit for the files stored in the storage location by the file storage unit. Calculate the suitability of the above storage location for
The storage location candidate extraction unit extracts a storage location candidate for the data input by the data input unit from the plurality of storage locations based on the fitness calculated by the storage location suitability calculation unit. The similarity search apparatus according to claim 1.   The degree of conformity calculation unit calculates the sum or average of the similarities calculated by the similarity degree calculation unit for the files stored in the storage location to obtain the conformity of the storage location. 3. The similarity search device according to 2. The similar search device further includes a similar file extraction unit,
The similar file extraction unit extracts a file similar to the data input by the data input unit from the plurality of files based on the similarity calculated by the similarity calculation unit. The similarity search device according to any one of claims 1 to 3.   The similarity calculation unit includes, among the headwords extracted from the file stored in the file storage unit, the number of headwords matching the headword extracted by the headword extraction unit from the data input by the data input unit or 5. The similarity search apparatus according to claim 1, wherein a ratio is calculated to obtain a similarity between the file and the data.   A computer program that, when executed by a computer, causes the computer to function as the similarity search device according to any one of claims 1 to 5. A storage device stores a plurality of files,
The storage device stores the headword extracted from the file for each stored file,
The input device inputs the data,
The processing device extracts headwords from the data input by the input device,
The processing device searches for a headword that matches the extracted headword from the headwords stored in the storage device,
A similarity search method, wherein the processing device calculates a similarity between each file stored in the storage device and data input by the input device based on the searched entry word.

Images