JP2004220513A - Information retrieval device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information retrieval device appropriately retrieving information according to a specified retrieval requirement. <P>SOLUTION: This information retrieval device selects unnecessary words corresponding to the language of a retrieval requirement sentence (step S11), refers a retrieval method in a retrieval method specificating part, when a concept retrieval is specified, selects unnecessary words for the concept retrieval and, when a logical retrieval is specified, selects the unnecessary words for the logic retrieval (step S12) Then, this device refers a retrieval object field stored in a buffer memory in a retrieval object field detecting part, and selects the unnecessary words agreed with the specified retrieval object field (step S13). When the retrieval object field is unspecified or there is no unnecessary word corresponding to the agreed field, this device refers unnecessary word selection specification information stored in an unnecessary word selection specification holding part, when the unnecessary words to be used is specified, selects the unnecessary words shown in the unnecessary word selection specification holding part irrelevant to the result of the unnecessary word selection in the above steps (step S14). <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an information search device that manages a plurality of types of information and searches for desired information from the managed information.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, as an information search device for searching for information, for example, image information accompanied by a content description of a document or a character content, a word or sentence representing desired information is input and matched with the input word or sentence. 2. Description of the Related Art There is known an information search apparatus that applies a technique called full-text search, which holds words or sentences to be searched and obtains accumulated information.
[0003]
There is also known an information search device that obtains not only information that holds a word or sentence that matches an input word or sentence, but also information that holds a word or sentence that is determined to have a similar concept to the input word or sentence. ing.
[0004]
In such an information search device, in order to frequently appear in any information, the search efficiency is improved by excluding, from the search target, universal words that do not contribute to the identification of the information. A word called an unnecessary word to be excluded from the search target is stored in the apparatus.
[0005]
[Problems to be solved by the invention]
However, in the above-described conventional information search device, unnecessary words are always excluded from search targets, and the problem that words to be searched are defined as unnecessary words depending on the operation purpose of the operator is not searched. there were.
[0006]
To solve such problems, there is an information search device that does not use unnecessary words at the time of search, that is, all search words are searched. However, it is not suitable for the operation purpose of the operator. There is a problem that a large number of uncommon words are searched for in large quantities.
[0007]
For example, the English pronoun “it” has little inherent meaning in itself and frequently appears regardless of stored information. Therefore, “it” appearing in a search request sentence is an unnecessary word from a search target. You want to exclude them. However, "it" as an English pronoun rarely appears in Japanese documents, and it is said that "IT" of "IT Revolution" is not searched for by making "it" an unnecessary word. It causes problems.
[0008]
Also, in a concept search for searching for a document having a concept similar to the content of a search request sentence, there is polysemy and it is not possible to determine the meaning of the word. A word that does not contribute to the term is regarded as an unnecessary word. However, in a logical search in which a document in which a search term appears is determined as true (appears) or false (does not appear) in binary, a word defined as an unnecessary word in a conceptual search is used. There is a problem that the words to be unnecessary words differ depending on the search method, for example, there are cases where the words should be searched for.
[0009]
The present invention has been made in view of these points, and an object of the present invention is to provide an information search device capable of performing an appropriate information search according to a specified search request.
[0010]
[Means for Solving the Problems]
In order to achieve the above object, an information retrieval apparatus according to claim 1 includes an input unit that inputs a search word or a search sentence, and at least character data based on the search word or the search sentence input by the input unit. A search unit that searches for partial information from a plurality of pieces of information, a storage unit that stores a plurality of types of unnecessary word groups including a plurality of words or character strings that are not used in the search by the search unit, and the stored plurality of types. Selecting means for selecting at least a part of unnecessary word groups from the unnecessary word groups according to a predetermined condition; and unnecessary information included in the character data group selected by the selecting means at the time of searching for information by the searching means. Exclusion means for excluding a word from the search word.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0012]
FIG. 1 is a block diagram showing a schematic configuration of an information search device according to one embodiment of the present invention.
[0013]
In FIG. 1, a microprocessor (CPU) 11 performs an operation and a logical decision for information retrieval, and connects to those buses AB, CB, and DB via an address bus AB, a control bus CB, and a data bus DB. Control the respective components 12 to 19 performed.
[0014]
The address bus AB transfers an address signal indicating each of the components 12 to 19 to be controlled by the CPU 11. The control bus CB transfers and applies control signals of the components 12 to 19 to be controlled by the CPU 11. The data bus DB transfers data among the components 11 to 19.
[0015]
The ROM 12 is a fixed read-only memory (read only memory), and stores control program codes such as processing programs executed by the CPU 11.
[0016]
The RAM 13 is a writable random access memory composed of 16 bits per word, and is used for temporarily storing various data from each component. Further, on the RAM 13, a search word holding unit 202, an unnecessary word selection designation holding unit 204, a search concept vector holding unit 212, and a search result holding unit 214, which will be described later with reference to FIG.
[0017]
The DISK 14 is an external memory, and stores an unnecessary word 207, a concept dictionary 211, a stored document concept vector holding unit 216, a stored document 218, and a word index 219, which will be described later with reference to FIG. Also, a search request input processing unit 201, an unnecessary word selection designation processing unit 203, a search method designation processing unit 205, an unnecessary word discrimination processing unit 206, a search target field detection processing unit 208, a search word selection process, which will be described later with reference to FIG. A program code for executing the processing units of the unit 209, the search concept vector creation processing unit 210, the concept search processing unit 213, the logic search processing unit 215, and the concept vector creation processing unit 217 is also stored.
[0018]
As a storage medium for storing these data and programs, a ROM, a floppy (registered trademark) disk, a CD-ROM, a DVD-ROM, a memory card, a magneto-optical disk, or the like can be used.
[0019]
The KB 15 is a keyboard, and includes various function keys such as an alphabet key, a hiragana key, a katakana key, a character symbol input key such as a period, a search key for instructing a search, and a cursor movement key for instructing a cursor movement. ing.
[0020]
The VRAM 16 is a video memory for display, and stores a pattern of data to be displayed.
[0021]
The CRTC 17 is a CRT controller and plays a role of displaying the contents stored in the VRAM 16 on the CRT 18.
[0022]
The CRT 18 is a cathode ray tube, and the display pattern of the dot configuration and the display of the cursor on the CRT 18 are controlled by the CRTC 17. Note that a display device such as a liquid crystal panel may be employed instead of the CRT.
[0023]
The NIC 19 is a network controller and has a role of connecting to a network such as Ethernet (registered trademark).
[0024]
The information retrieval device configured as described above operates according to various inputs from the KB 15 and various inputs via the network supplied from the NIC 19, and when there is an input from the KB 15 or the NIC 19, first, an interrupt signal is generated. Sent to CPU 11. In response, the CPU 11 reads various control signals stored in the disk 14 and starts various controls according to the control signals.
[0025]
FIG. 2 is a block diagram illustrating a functional configuration of the information search device according to the present embodiment.
[0026]
In the figure, a search request input processing unit 201 inputs a requirement (search sentence or search word) relating to a desired search target.
[0027]
The search term holding unit 202 stores the search term input by the search request input processing unit 201.
[0028]
The unnecessary word selection / designation processing unit 203 is for the user to specify a search unnecessary word, and the unnecessary word selection / designation holding unit 204 is a user's selection result of the unnecessary word specified by the unnecessary word selection / designation processing unit 203. Is stored by, for example, a configuration described later with reference to FIG.
[0029]
The search method designation processing unit 205 selects a document search method.
[0030]
The unnecessary word discrimination processing unit 206 discriminates and selects an unnecessary word to be applied from a plurality of unnecessary words.
[0031]
The unnecessary word storage unit 207 stores a plurality of unnecessary words defined according to the purpose.
[0032]
The search target field detection processing unit 208 detects and determines a field to be searched.
[0033]
The search word selection processing unit 209 selects a search target word from the search words stored in the search word holding unit 202 based on the unnecessary word selected by the unnecessary word determination processing unit 206.
[0034]
The search concept vector creation processing unit 210 creates a search concept vector by referring to the concept dictionary 211 based on the search word selected by the search word selection processing unit 209.
[0035]
The concept dictionary 211 describes and stores the semantic characteristics of the word serving as a headline.
[0036]
The search concept vector holding unit 212 stores the search concept vector created by the search concept vector creation processing unit 210.
[0037]
The concept search processing unit 213 performs a concept search.
[0038]
The logical search processing unit 215 performs a logical search.
[0039]
The search result holding unit 214 stores the processing results of the concept search processing unit 213 and the logical search processing unit 215.
[0040]
The concept vector creation processing unit 217 creates a concept vector for the registered document 220.
[0041]
The stored document concept vector holding unit 216 stores the concept vector created by the concept vector creation processing unit 217 in association with the registered document 220.
[0042]
The word index 219 stores an index of a word appearing in the registered document 220 created by the concept vector creation processing unit 217.
[0043]
The stored document 218 stores the registered document 220.
[0044]
FIG. 3 is a diagram illustrating a display example of the operation panel when the search request input processing unit 201 indicates a search request sentence or a search request word.
[0045]
In the figure, a display window 301 is for performing a search request input operation.
[0046]
An area 302 is a search sentence input area for inputting a sentence or a word as a search request.
[0047]
The sentence 303 indicates the search request sentence being input, and in the illustrated example, “mobile device market trend” is input.
[0048]
The cursor 304 indicates an input position in the search sentence input area 302.
[0049]
Buttons 305 and 306 are radio buttons for specifying a search method, and select one of them. In the illustrated example, “concept search” is selected.
[0050]
An area 307 is an area specification input area for specifying a target field of a document to be searched, and selects one from a list of fields displayed in a pull-down display. If no particular field is specified, “unspecified” is selected as the default.
[0051]
A button 308 is a search execution button for specifying execution of a search process. When the search execution button 308 is pressed, the search method specified by the radio button 305 or 306 is used for the search request sentence input in the search sentence input area 302. Is executed based on the search processing.
[0052]
A button 309 is a cancel button for designating termination or cancellation of the search processing. When the cancel button 309 is pressed, the search processing is immediately terminated, and the display window 301 is closed and terminated.
[0053]
An area 310 is a search result display area for displaying the result of performing the search processing by pressing the search button 308. In the illustrated example, nothing is displayed since the search processing has not been performed.
[0054]
FIG. 4 shows a display example of a search result when a search request sentence or a search request word for performing a concept search or a logical search is specified by an operator in the search request input processing unit 201 and a search process is executed. FIG.
[0055]
3, an area 401 is a display area of a search result, like the area 310 of FIG.
[0056]
The area 402 is a rank display area indicating the order of the search result. Search results are ranked in order of similarity to the search request and are displayed in rank order. In the illustrated example, search results of ranks 25 to 30 are displayed.
[0057]
An area 403 is an area for displaying the title of the searched document, and an area 404 is an area for displaying the file name of the document.
[0058]
An area 405 is an area for displaying the contents of the searched document that can be understood. In the document content display area 405, it is possible to display an abstract sentence given in advance as a bibliographic attribute of the document, an abstract automatically summarized from the document, or a general sentence extracted from a part of the document as a meaning. it can.
[0059]
When the bar 406 cannot be displayed in the search result display area 401, the search result is partially displayed in the display area 401 while displaying other parts that are not displayed, and the display position is designated. This is an elevator bar used in the same type of window display device.
[0060]
The display state shown in FIG. 4 displays a search result obtained by performing a conceptual search on “market trend of mobile device” input as the search sentence 303.
[0061]
FIG. 5 is a diagram showing a configuration of the concept dictionary 211. The concept dictionary 211 expresses the concept of a word as a multidimensional vector having universal semantic weights as elements.
[0062]
In the figure, an area 501 stores a word serving as a heading of the concept dictionary 211.
[0063]
An area 502 stores a subscript representing each element of a semantic vector expressed in 256 dimensions for the headword 501.
[0064]
An area 503 stores the weight of each element of the semantic vector, the weight takes a real number between 0 and 1, and is normalized and stored so that the size of the semantic vector becomes 1.
[0065]
A universal semantic element that is an element of a multidimensional vector that forms the concept dictionary 211 is a label that represents a single semantic concept, and includes, for example, words such as “this, that, that, which”. Words such as "concept of instruction", "class, grade, grade, rank, rank, inferiority, superiority, honor" and so on are included in words such as "concept of grade", "change, transformation, innovation, rise". Words such as "concept of change", "cooperation, greeting, unity, handshake, friendship, diplomacy, friendship" are included, "concept of fellowship", "animal, mammal, penguin, dog, human, We use independent and universal semantics that do not depend on a specific word, such as "concept of living things", which is included in words such as "goldfish". In FIG. 5, 256 types of semantics are used to form a 256-dimensional concept expression vector.
[0066]
FIG. 6 is a diagram showing a configuration of the word index 219. The word index 219 is a table that stores the appearance frequency in the document for all words that appear in the registered document.
[0067]
In the figure, the first column information 601 of the table is a document ID for uniquely identifying a registered document. The second to n-th column information 602 of the table are suffixes indicating each word shown in FIG. The last row 603 of the table stores the sum total of the number of occurrences of each word. In FIG. 6, the document with the document ID “00146” indicates that the word “market” indicated by the subscript “1256” appears in the document 12 times.
[0068]
FIG. 7 is a diagram showing a configuration of a correspondence table in which the correspondence between words and subscripts of the word index table is stored in the word index 219.
[0069]
In the figure, the correspondence table stores a word 701 and a unique word index 702 associated with each other. For example, it is indicated that the word index of the word “market” is “1256”.
[0070]
Next, the configuration of the unnecessary word 207 will be described with reference to FIGS. The unnecessary word 207 stores a plurality of unnecessary words.
[0071]
FIG. 8 exemplifies unnecessary words to be referred to at the time of executing a logical search for Japanese as a search target. In the figure, a line having a pound sign (#) at the beginning of the line is a comment line, and unnecessary word definition data is data of the second and subsequent lines, which are arranged in the order of character codes.
[0072]
FIG. 9 exemplifies unnecessary words to be referred to at the time of executing a logical search for English as a search target.
[0073]
8, as in FIG. 8, a sharp symbol at the beginning of a line indicates a comment line, and unnecessary words in English are arranged in the order of character codes.
[0074]
FIG. 10 exemplifies unnecessary words to be referred to at the time of executing a concept search for English as a search target. This unnecessary word has the same configuration as that of the unnecessary word in FIGS. 8 and 9 and exemplifies the unnecessary word for concept search reference in the order of the character code.
[0075]
Next, the configuration of the unnecessary word selection designation holding unit 204 will be described with reference to FIG.
[0076]
The unnecessary word selection designation holding unit 204 is configured such that the operator directly specifies an unnecessary word to be used at the time of performing a search from a plurality of unnecessary words stored in the unnecessary word 207, and has a sharp symbol at the beginning in FIG. The line is a comment line and the data is skipped. An unnecessary word file having a file path associated with the [Stop Word List] section of the unnecessary word label that is ON defined in the [Stop Word Selection] section is used. In the example of FIG. 11, unnecessary words having the labels of STOP1 and STOP2 for which ON is specified in the [Stop Word Selection] section are specified. In the [Stop Word List] section, C: \ JBOOL. dat, C: {EBOOL. Unnecessary words specified by the dat file path are used in the search processing.
[0077]
Next, a document search process executed in the present embodiment will be described with reference to FIG.
[0078]
FIG. 12 is a flowchart illustrating a procedure of a document search process performed by the information search device according to the present embodiment.
[0079]
In FIG. 5, in step S1, a search request input process is performed by a processing module that operates the search request input processing unit 201. In the search request input process, the search request sentence 303 input to the search request sentence input area 302 is stored in the search word holding unit 202, and the search target field selected by the search target field designation 307 is searched for by the search target field. It is stored in a buffer memory (not shown) in the field detection processing unit 208. The search method specified by the radio button 305 or 306 is stored in a buffer memory (not shown) in the search method specification processing unit 205.
[0080]
In step S2, an unnecessary word to be selected at the time of executing a search is determined based on conditions such as a search method, an unnecessary word selection designation, a search target field, and a search target language. The details of the unnecessary word selection processing will be described later with reference to FIG.
[0081]
In step S3, a search method is determined. In the search request input processing in step S1, the search method stored in the buffer memory in the search method specification processing unit 205 is referred to. If the concept search has been specified, the process proceeds to step S4. If the logical search has been specified. For example, the process branches to step S5.
[0082]
In step S4, a concept search process for searching for a document having a similar concept represented by the document content is performed according to the search request input in the search request input process step S1. The details of the concept search process will be described later with reference to FIG.
[0083]
In step S5, according to the search request input in the search request input processing step S1, a logical search process is performed to search for a document in which a word appearing in the search request sentence appears by referring to the word index 219. The details of the logical search process will be described later with reference to FIG.
[0084]
In step S6, the search result searched in step S4 or step S5 is retrieved from the search result holding unit 214 and displayed. This process is a known process widely performed in the same type of information search device.
[0085]
FIG. 13 is a flowchart showing a detailed procedure of the unnecessary word selection processing in step S2.
[0086]
In the figure, in step S11, unnecessary words are selected according to the language type of the search request language. The description language of the search request sentence stored in the search word holding unit 202 is identified, and an unnecessary word corresponding to the language is selected. For example, if the search request sentence is described in English, an unnecessary word for English is selected, and if the search request sentence is described in Japanese, an unnecessary word for Japanese is selected. The description language identification of the search request sentence is not shown, but a method of specifying a language in advance, a method of selecting a language type of a dictionary having the highest dictionary matching rate of words obtained from the search request sentence, and forming a search request sentence For example, a method of estimating a language from the distribution of character codes to be used can be selected.
[0087]
Next, in step S12, unnecessary words are selected by a search method. Referring to the search method stored in the buffer memory in the search method specification processing unit 205 in step S1 of FIG. 12, if a concept search is specified, an unnecessary word for the concept search is selected, and a logical search is specified. In this case, an unnecessary word for a logical search is selected.
[0088]
Next, in step S13, unnecessary words are selected according to the search target field. In step S1, the search target field stored in the buffer memory in the search target field detection processing unit 208 is referred to, and an unnecessary word that matches the specified search target field is selected. When the search target field is not specified or when there is no unnecessary word corresponding to the matching field, the process proceeds to the next step.
[0089]
Next, in step S14, unnecessary words are selected by specifying unnecessary words. By referring to the unnecessary word selection designation information stored in the unnecessary word selection designation holding unit 204, if an unnecessary word to be used is designated, regardless of the result of the unnecessary word selection from step S11 to step S13, An unnecessary word shown in the unnecessary word selection designation holding unit 204 is selected.
[0090]
FIG. 14 is a flowchart showing a detailed procedure of the concept search process in step S4 of FIG.
[0091]
In the figure, in step S21, the search sentence stored in the search word holding unit 202 is extracted and divided into words. For dividing the search sentence into words, a method known as morphological analysis processing is applied.
[0092]
Next, in step S22, referring to the unnecessary word selected in step S2 of FIG. 12, it is searched whether the word extracted in step S21 is included in the unnecessary word. If it is not an unnecessary word, the process immediately branches to step S24, and if it is an unnecessary word, the process proceeds to step S23.
[0093]
In step S23, it is determined whether or not the language type of the unnecessary word matched in step S22 matches the language type of the search word extracted in step S21. This determination is for applying an appropriate unnecessary word when different languages are mixed. For example, the first word “It” extracted from the search request sentence “It is IT revolution” is If it is an English word and “It” is defined as an unnecessary word for English, this search word should be excluded as an unnecessary word, but “IT” of “IT revolution” having the same character string Is a part of a Japanese word, and the linguistic information does not match, so that it can be adopted as a search word. Note that uppercase and lowercase letters are normalized and identified.
[0094]
In step S24, a search is made to see if there is any entry that matches the entry word 501 in the concept dictionary 211. If there is a headword 501 of the concept dictionary 122 that matches the search word, the corresponding concept vector data is extracted from the concept dictionary 211.
[0095]
In step S25, the retrieved concept vector is created by adding the component values of the components of the acquired concept vector data. In addition, all the dimension elements of the search concept vector are initialized to “0” in advance.
[0096]
In step S26, it is determined whether or not all the search terms of the search term holding unit 202 have been processed. When the processing of all of the search terms has been completed, the search concept vector data is changed to a sum of squares of each element of “1”. After normalization is performed, the result is stored in the search concept vector holding unit 212, and the process branches to step S27. If there is an unprocessed search word, the process returns to step S21.
[0097]
In step S27, the stored document concept vector is acquired from the stored document concept vector storage unit 216, and the concept similarity between the stored document concept vector and the search request concept vector stored in the search concept vector storage unit 212 is calculated. The concept similarity can be calculated by the cosine measure of both vector data. The calculated concept similarity is temporarily stored in a buffer memory (not shown) in association with the document ID of the stored document.
[0098]
After finishing the processing of step S27 for all the stored documents, the process proceeds to step S28, where the search results are sorted in descending order of the concept similarity temporarily stored in step S27, and stored in the search result holding unit 214. The concept search process ends.
[0099]
FIG. 15 is a flowchart showing a detailed procedure of the logic search process in step S5 of FIG.
[0100]
In the figure, in step S31, the search sentence stored in the search word holding unit 202 is extracted and divided into words. For dividing the search sentence into words, a method known as morphological analysis processing is applied.
[0101]
Next, in step S32, the unnecessary word selected in step S2 in FIG. 12 is referred to, and it is searched whether the word extracted in step S31 is included in the unnecessary word. If it is not an unnecessary word, the process immediately branches to step S34, and if it is an unnecessary word, the process proceeds to step S33.
[0102]
In step S33, it is determined whether or not the language type of the unnecessary word matched in step S32 matches the language type of the search word extracted in step S31. This determination is for applying an appropriate unnecessary word when different languages are mixed. For example, the first word “It” extracted from the search request sentence “It is IT revolution” is If it is an English word and “It” is defined as an unnecessary word for English, this search word should be excluded as an unnecessary word, but “IT” of “IT revolution” having the same character string Is a part of a Japanese word, and the linguistic information does not match, so that it can be adopted as a search word. Note that uppercase and lowercase letters are normalized and identified.
[0103]
In step S34, collation with the word index 219 is performed, and a search is performed to determine whether there is a document including the search term extracted in step S31. As a result of collation with the word index 219, the information of the document ID and the word appearance frequency of the searched document is temporarily stored in a buffer (not shown).
[0104]
In step S35, it is determined whether or not all the search terms in the search term holding unit 202 have been processed, and when all the search terms have been processed, the flow branches to step S36. If there is an unprocessed search word, the process returns to step S31.
[0105]
After the processing for all the stored documents is completed, in step S36, the search results are sorted in descending order of the word frequency temporarily stored in step S34, stored in the search result holding unit 214, and the logical search processing ends. I do.
[0106]
Next, a document registration process executed in the present embodiment will be described with reference to FIG.
[0107]
FIG. 16 is a flowchart illustrating a procedure of a document registration process performed by the information search device according to the present embodiment.
[0108]
In the figure, in step S41, words are extracted from the registered document 220. For the extraction of words, a method generally used for morphological analysis is applied.
[0109]
Next, in step S42, registration in the word index 219 is performed. If the word matches the word 701 in the word index table, the corresponding word ID is obtained, and the appearance frequency in the target registered document is stored in the column information using the word ID as an index. If the corresponding word does not match the word 701 in the word index table, the word is added to the word index table, a new unique word ID is allocated, and the allocated word ID is used as a new word index, and column information is added. For a stored document that has already been registered, “0” is stored as the frequency, and for a target registered document, the appearance frequency is stored.
[0110]
Next, in step S43, the concept dictionary 211 is searched for a headword 501 that matches the word, and if a matching headword exists, the corresponding concept vector data is extracted.
[0111]
In step S44, the concept vector data extracted in step S43 is multiplied by a weight corresponding to the frequency, and is added to the document concept vector data.
[0112]
In step S45, it is determined whether or not processing has been completed for all words in the registered document. If there are unprocessed words, the process returns to step S42. If processing has been completed for all words, the process branches to step S46.
[0113]
In step S46, the document concept vector data is normalized so that the sum of the squares of the vector elements becomes "1", registered in the stored document concept vector holding unit 216, and the document registration process ends.
[0114]
Although the present embodiment has been described using document information as a search target, multimedia information such as image information, video information, and program content description information accompanied by content description metadata other than the document information can also be used. The present invention can be applied by extracting the information feature amount from the described sentence information and obtaining a similarity measure of the information feature amount.
[0115]
Further, in the present embodiment, as a concept vector for concept search, a word attribute concept vector is stored in the concept dictionary 211, and is configured by a composite vector. It is also possible to configure a concept vector for a concept search by a word vector having a weight calculated from information or a word frequency and an inverse document frequency as an element.
[0116]
Further, in the present embodiment, it is determined whether or not a word appearing in the search request sentence is an unnecessary word, and if the word is unnecessary, it is excluded from the search word. In the word index and document concept vector creation process, whether the word extracted in step S41 is the selected unnecessary word is performed in the same manner as in steps S22 and S23 in FIG. Only when it is determined, the process proceeds to step S42. When the word is an unnecessary word, the process returns to step S41 without going through the steps of creating the word index and the document concept vector. An advantage that the size of the holding portion 216 can be reduced can also be provided.
[0117]
In the present embodiment, the stored document 220 to be searched, the stored document concept vector holding unit 216, the word index 219, and the concept dictionary 211 are described as being arranged in the DISK 14 constituting a single device. It is also possible to disperse these components in different devices and perform processing on a network via the NIC 19.
[0118]
The present invention can be applied to a system including a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), but may be applied to a device including one device (for example, a copying machine, a facsimile machine, etc.) May be applied.
[0119]
A storage medium storing a program code of software for realizing the functions of the above-described embodiments is supplied to a system or an apparatus, and a computer (or CPU or MPU) of the system or the apparatus stores the program stored in the storage medium. It goes without saying that the object of the present invention is also achieved by reading and executing the code.
[0120]
In this case, the program code itself read from the storage medium implements the novel function of the present invention, and the storage medium storing the program code constitutes the present invention.
[0121]
Examples of a storage medium for supplying the program code include a flexible disk, a hard disk, a magneto-optical disk, a CD-ROM, a CD-R, a CD-RW, a DVD-ROM, a DVD-RAM, a DVD-RW, a DVD + RW, and a magnetic disk. A tape, a nonvolatile memory card, a ROM, or the like can be used. Further, the program code may be supplied from a server computer via a communication network.
[0122]
In addition, the functions of the above-described embodiments are implemented when the computer executes the readout program codes, and the OS or the like running on the computer performs the actual processing based on the instructions of the program codes. It goes without saying that a part or all of the above is performed, and the function of the above-described embodiment is realized by the processing.
[0123]
Further, after the program code read from the storage medium is written into a memory provided on a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.
[0124]
Hereinafter, examples of embodiments of the present invention will be listed.
[0125]
(Embodiment 1) Input means for inputting a search word or a search sentence,
Searching means for searching for partial information from a plurality of information including at least character data based on a search word or a search sentence input by the input means;
Storage means for storing a plurality of types of unnecessary word groups consisting of a plurality of words or character strings not used in the search by the search means;
Selecting means for selecting at least some of the unnecessary word groups from the stored plurality of types of unnecessary word groups according to a predetermined condition;
An exclusion unit for excluding unnecessary words included in the character data group selected by the selection unit from the search word when searching for information by the search unit;
An information retrieval device, comprising:
[0126]
(Embodiment 2) The storage unit stores the unnecessary word groups in association with each of the plurality of pieces of language information,
The predetermined condition is to select an unnecessary word group corresponding to the search word input by the input unit or the linguistic information of the search sentence.
The information retrieval device according to the first embodiment, characterized in that:
[0127]
(Embodiment 3) The search means searches for information using any one of a plurality of search methods selected by a user,
The storage unit stores the unnecessary word groups in association with each of the plurality of search methods,
The predetermined condition is to select an unnecessary word group corresponding to the search method selected as the search means.
The information retrieval device according to the first embodiment, characterized in that:
[0128]
(Embodiment 4) The plurality of search methods include a logical search unit that searches for information including character data corresponding to a search word or search sentence input by the input unit. Described information retrieval device.
[0129]
(Fifth Embodiment) The plurality of search methods may include information including character data corresponding to a similar concept, based on a similarity evaluation scale representing a concept feature of a search word or search sentence input by the input unit. The information search apparatus according to the third embodiment, further comprising a concept search unit for searching for a search.
[0130]
(Embodiment 6) The input means can input a search field to be searched,
The storage unit stores the unnecessary word groups in association with each of the plurality of search fields,
The predetermined condition is that an unnecessary word group corresponding to the search field input by the input unit is selected.
The information retrieval device according to the first embodiment, characterized in that:
[0131]
(Embodiment 7) The information search device according to embodiment 1, wherein the predetermined condition is an instruction from a user.
[0132]
(Eighth Embodiment) A retrieval step of retrieving partial information from a plurality of pieces of information including at least character data based on a search word or a search sentence input by an input means for inputting a search word or a search sentence,
According to a predetermined condition, at least a part of unnecessary word groups is stored from a plurality of types of unnecessary word groups stored in a storage unit that stores a plurality of types of unnecessary word groups including a plurality of words or character strings not used in the search in the search step. A selection step of selecting a word group;
An exclusion step of excluding unnecessary words included in the character data group selected by the selection step from the search word when searching for information by the search step;
An information search method characterized by having:
[0133]
(Embodiment 9) A search procedure for searching for partial information from a plurality of pieces of information including at least character data based on a search word or a search sentence input by an input means for inputting a search word or a search sentence,
According to a predetermined condition, at least a part of unnecessary word groups is stored from a plurality of types of unnecessary word groups stored in a storage unit that stores a plurality of types of unnecessary word groups including a plurality of words or character strings that are not used in the search by the search procedure. A selection procedure for selecting a word group,
An exclusion step of excluding unnecessary words included in the character data group selected by the selection step from the search word when searching for information by the search step;
A program for causing a computer to execute.
[0134]
【The invention's effect】
As described above, according to the present invention, a plurality of types of unnecessary words are held and unnecessary words are selected in accordance with a specified search request, so that information search with excellent operability and high search accuracy can be performed. Can be.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a schematic configuration of an information search device according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a functional configuration of the information search device of FIG. 1;
FIG. 3 is a diagram showing a display example of an operation panel when a search request sentence and a search method are input in the search request input processing unit in FIG. 2;
4 is a diagram showing a display example of a search request sentence and a search result corresponding to a search method in the search request input processing unit of FIG. 2;
FIG. 5 is a diagram showing a configuration of a concept dictionary of FIG. 2;
FIG. 6 is a diagram illustrating a configuration of a word index in FIG. 2;
FIG. 7 is a diagram showing the correspondence between word IDs and words in the word index of FIG. 6;
FIG. 8 is a diagram showing search-unnecessary words in Japanese logic search.
FIG. 9 is a diagram showing search-unnecessary words in an English logical search.
FIG. 10 is a diagram showing search unnecessary words in an English concept search.
FIG. 11 is a diagram showing a definition of search-unnecessary word selection designation.
FIG. 12 is a flowchart illustrating a procedure of a document search process executed by the information search device of FIG. 1;
FIG. 13 is a flowchart showing a detailed procedure of an unnecessary word selecting process in FIG. 12;
FIG. 14 is a flowchart showing a detailed procedure of a concept search process of FIG. 12;
FIG. 15 is a flowchart illustrating a detailed procedure of a logical search process in FIG. 12;
FIG. 16 is a flowchart showing a procedure of index information creation processing for document search executed by the information search device of FIG. 1;
[Explanation of symbols]
11 CPU
12 ROM
13 RAM
14 DISK
15 KB
16 VRAM
17 CRTC
18 CRT
19 NIC
201 search request input processing unit
202 Search Term Holder
203 Unwanted word selection designation processing unit
204 Unnecessary word selection designation holding unit
205 Search method specification processing unit
206 Unnecessary word discrimination processing unit
207 unnecessary words
208 Search target field detection processing unit
209 Search word selection processing unit
210 Search Concept Vector Creation Processing Unit
211 Concept Dictionary
212 Search concept vector storage
213 Concept Search Processing Unit
214 search result storage
215 Logical search processing unit
216 Stored Document Concept Vector Holder
217 Concept vector creation processing unit
218 Stored document
219 Word Index
220 Registration Document

Claims (1)

An input means for inputting a search word or a search sentence,
Searching means for searching for partial information from a plurality of information including at least character data based on a search word or a search sentence input by the input means;
Storage means for storing a plurality of types of unnecessary word groups consisting of a plurality of words or character strings not used in the search by the search means;
Selecting means for selecting at least some of the unnecessary word groups from the stored plurality of types of unnecessary word groups according to a predetermined condition;
An information retrieval apparatus, comprising: an exclusion unit configured to exclude an unnecessary word included in a character data group selected by the selection unit from the search word when searching for information by the search unit.

Images