JP5794001B2 - Information search method, information search device, and information search program - Google Patents

Description

  The present invention relates to a search method and a search device for extracting information related to an input keyword from a search target.

  There is known a search system that performs a search in consideration of the importance of each keyword when a plurality of search keywords are input (see, for example, Patent Document 1).

JP-A-5-158991

  When a plurality of search keywords include modifiers, a search result preferable for the searcher cannot be obtained even if a search is performed in consideration of the importance of the keywords, without considering the modifier relationship (how to engage).

Information search method according to the present invention includes a first step of the computer accepts an input of a keyword, computer, based on an input form or input sequence of the plurality of keywords when multiple keywords are entered in a first step, a second step of determining the priority of a modified relationship by keyword multiple, computer, based on the determined priority, and a third step of outputting searching for information related to a plurality of keywords from the search with Ru.
Information retrieval apparatus according to the present invention includes a receiving unit that receives an input of a keyword, the plurality of keywords are input to the reception unit, based on an input form or order of inputting multiple keywords, modification by keyword multiple a determination unit for determining a priority of relations, based on Sakido Yu, Ru and an output unit for outputting searching for information related to a plurality of keywords from the search target.
The information search program according to the present invention includes a first step of accepting keyword input to a computer, and when a plurality of keywords are input in the first step, a plurality of keywords are input based on an input form or input order of the plurality of keywords. A second step of determining the priority of the modification relationship by keyword and a third step of searching and outputting information related to a plurality of keywords from the search target based on the determined priority are executed.

  According to the present invention, when a search is performed by inputting a plurality of keywords, it is possible to increase the possibility of obtaining a detection result preferable for the searcher.

1 is a schematic configuration diagram of a system in one embodiment. The figure which shows the structure of a server computer. The figure which shows the example of an input of the search screen in user PC. The figure which shows the example of a result display when the search is instruct | indicated with the input value of FIG. The figure which shows an example of the search processing procedure in a server computer. The figure which shows the example which inputs a keyword into an individual input part.

An embodiment when the present invention is applied to an image search system on the Web will be described based on the drawings.
FIG. 1 shows a schematic configuration of an image search system in the present embodiment. A person who performs an image search (hereinafter referred to as a user) accesses the server computer 100 of the image search site from his / her personal computer or the like (hereinafter referred to as a user PC) 50 via the network 60 and searches a search keyword (hereinafter simply referred to as a keyword). To search for images. In the present embodiment, the embodiment of the present invention will be described by taking the case where the present invention is applied to an image search system on the Web as an example. However, the present invention is not limited to this.

FIG. 2 is a block diagram showing the configuration of the server computer 100. The server computer 100 includes a main body having a CPU 11, a RAM 12, a disk interface 13, a display controller 14, a CPU bus 15, a serial bus 16, a serial bus controller 17, etc., and a keyboard 21 and a mouse 22 connected via the serial bus 16. And the like, a storage unit 31 connected via the disk interface 13, and a display 32 as a display unit. The storage unit 31 stores an image database for storing image data to be searched and a search program (search engine) for searching for an image from the image database.
The user PC 50 also has substantially the same hardware configuration as the server computer 100.

  Image data to be searched (hereinafter simply referred to as an image) varies depending on the form of the image search site. For example, a wide range of images posted on a Web page may be a search target. In a photo album site, an image posting site, or the like, only images posted (uploaded) on the site are search targets. In the latter case, there may be a service in which only images posted by the user in the past are targeted for search.

  A CPU (hereinafter referred to as server CPU) 11 of the server computer 100 performs a search process according to a search program. The search program has a function of searching (extracting) an image related to the input keyword (more precisely, an image associated with information related to the keyword) from the image database and outputting the image. For example, if an image posted on a Web page is a search target, the relevance between the text description on the page and the input keyword is determined, and if there is a relevance, the image is extracted. Is done. The criterion for determining the presence or absence of relevance depends on the algorithm of the search program.

  The image exists as an image file that can be handled by a computer, and various additional information is written in the header of each image file in the camera that captured the image file. It is also possible for the user to write some information into the image file as additional information after shooting. In the search, it is possible to perform the extraction by judging the relationship between the additional information and the keyword.

  Examples of the additional information include scene shooting mode information. When using a digital camera equipped with a scene shooting mode (“Portrait shooting mode”, “Landscape shooting mode”, “Night scene shooting mode”, “Cooking shooting mode”, etc.) When a person or camera sets a scene shooting mode according to the subject, the camera automatically sets shooting conditions suitable for the subject and performs shooting. After shooting, the camera writes scene shooting mode information indicating that the scene shooting mode is used as additional information in the image file. If this scene shooting mode information is used for the search, for example, if “portrait shooting mode” is described as the scene shooting mode information in the search target image file, the image data includes “person”, “person”, Can be extraction candidates for keywords such as “portrait”.

  Further, in a camera having a known face recognition function, when a face is recognized in image data, “person” or “portrait” is added as information when the camera is added, and particularly when a plurality of faces are recognized, “ By writing “group photo” or the like, the information can be used for the search as described above. Alternatively, a person in the image may be specified using the face recognition function, and the information may be retrieved by writing information on the specific person as the added information.

  In addition, it is also possible to identify the subject by analyzing the image data and determine the presence or absence of extraction by judging the relationship between the identification result and the keyword.

  The search also reflects the settings of various search options that can be set in advance by individual users. Search options include, for example, designation of AND search or OR search, setting of the upper limit value of the number of extracted images, designation of image size, designation of file format, and the like.

Hereinafter, the search procedure will be described by taking an example of searching for an image pasted on a Web page.
When the user PC 50 accesses the server computer 100, the server CPU 11 transmits screen display data to the user PC 50, and the user PC 50 receives this and displays a search screen on the display 32.

  FIG. 3 shows a display example of the search screen. A text box 71 for keyword input and a search button 72 are displayed on the display 32. In the text box 71, one or a plurality of keywords can be input using a keyboard. When the search button 72 is clicked after the keyword is input, a search instruction is transmitted to the server computer 100, and the server CPU 11 executes the search in response to this.

  In FIG. 3, three keywords “red”, “circle”, and “white” are entered separated by spaces. When the AND search is designated, the server CPU 11 extracts images related to all of “red”, “circle”, and “white”, and transmits the thumbnail data to the user PC 50. In the user PC 50, for example, as shown in FIG. 4, the transmitted thumbnail data is displayed as a list on the display 32 as thumbnail images SN.

  Here, in the conventional search method that does not consider the importance of keywords, the weights of “red”, “circle”, and “white” are the same, and an image in which all these characters are included in the explanatory text is displayed. Extract in order of increasing number of times or newest registration date. Also, when considering the importance of keywords as in Patent Document 1, for example, the search is performed by increasing the weight in the order of input, that is, “red”, “circle”, and “white”. An image related to a keyword with a large is preferentially extracted.

  However, in any of the above search methods, the search result is an image including a round object in red and white colors, an image including a red object, a round object, a white object (separate objects), a red object and a white object. An image including an object, an image including a red object and a white and round object, an image including an object with a red-white color and a round object, and the like are mixed with no particular priority. For this reason, if the user desires an image including “a red, round object and white object”, a desired search result is not obtained.

  Therefore, in this embodiment, when the keyword is a word that can be used as a modifier instead of simply weighting by importance, the priority of the modification relationship by the keyword is determined based on the importance, and the search is performed based on the priority. To do. For example, as shown in FIG. 3, when three words (all of which are modifiers) are input in the order of “red”, “circle”, and “white”, the server CPU 11 performs the input order, that is, “red”, “ It is recognized that the importance is high in the order of “circle” and “white”, and the priority of the keyword modification relationship is determined based on the recognition. The highest priority is “red, round and white”, followed by “red and round”. That is, “red and round”, which is a combination of “red” and “circle” with importance 1 and 2, is given priority over “red and white” and “round and white”.

  The server CPU 11 performs a search based on the priority of the keyword modification relationship. In other words, an image including a red, round, and white object is searched first, and then an image including a red, round object and a white object is searched preferentially. An image including a red object, a white and round object, and an image including a red object, a round object, and a white object have a low priority, and the extraction order is later.

  Here, when searching for an image including a red and round object and a white object, the server CPU 11 selects any one of “red circle”, “red circle”, “red circle”, “red circle”, “red circle”, and the like. For example, images in which the character string and the character string “white” are included in the explanatory text are extracted in the order of the highest display count or the newest registration date. Since there are few images that are red, round, and white, the search results corresponding to the input in FIG. 3 occupy a large number of images that include red, round, and white objects.

  If 3 words are entered in the order of “white”, “circle”, “red”, “white and round” has higher priority than “red and white” and “round and red” When “red”, “white”, and “circle” are input in this order, “red and white” has higher priority than “red and round” and “white and round”. When searching for an image including a red and white object and a round object, the server CPU 11 selects any one of the character strings such as “red and white”, “red and white pattern”, “red and white”, and the character string “circle”. Are extracted, for example, in descending order of display count or in order of new registration date.

  When a plurality of input keywords do not include a word that can be a modifier, the conventional AND search is performed. In this case, it is desirable to perform a search in consideration of the importance according to the keyword input order. In addition, when a plurality of keywords include a word that cannot be a modifier, a word that cannot be a modifier may be captured as a modifier. For example, if the keywords are entered in the order of “white”, “circle”, “red”, “dish”, “dish” is treated as a modified word, and an image containing a white round dish Search with priority over.

  FIG. 5 is a flowchart showing an example of search processing by the server CPU 11. This process is executed by the server CPU 11 in accordance with a search program stored in the storage unit 31 in advance.

  In step S1, a search start operation (mouse click of search button 72) is waited for by the user, and when the search start operation is confirmed, input of a keyword is accepted in step S2. In step S3, it is determined whether the search is an AND search or an OR search. If the OR search is set, an OR search based on the keyword is performed in step S4, and the process proceeds to step S8.

  On the other hand, if AND search is set, the importance of the keyword is recognized according to the keyword input order in step S5, and then in step S6, the priority of the modification relation by the keyword is based on the importance as described above. Determine the degree. In step S7, an AND search is performed based on the priority.

  In step S8, the image data (thumbnail data) extracted in step S4 or S7 is sequentially transmitted to the user PC 50, and then the process returns to step S1. The user PC 50 receives the thumbnail data transmitted from the server computer 100 and displays a list of thumbnail images on the display 32.

  Instead of the method of determining the importance in the above input order, the user indicates the importance before or after each keyword (for example, A, B, C... For example, if 1, 2, 3,... Are input in descending order of importance, the server CPU 11 may determine the importance of the keyword based on them. As an input example, when the input is “red (A), white (C), circle (B)”, a search result similar to the input example of FIG. 3 is obtained. Further, in the case of a format in which each keyword is input to each box as shown in FIG. 6, it may be determined that the weight input is higher as the keyword is input above.

  The image search has been described above, but the search target is not limited to image data. For example, the present invention can be applied to search for moving image data, music data, text data, or a Web page.

11 CPU
31 storage unit 50 user PC
60 network 71 text box 72 search button 100 server computer

Claims (6)

An information retrieval method executed by a computer,
A first step in which the CPU accepts an input of a keyword;
A second step in which, when a plurality of keywords are input in the first step, the CPU determines the priority of the modification relationship by the plurality of keywords based on the input form or input order of the plurality of keywords;
An information search method comprising: a third step in which the CPU searches and outputs information related to the plurality of keywords based on the determined priority.   The information search method according to claim 1, wherein the information to be searched includes image data. A reception unit that accepts keyword input;
When a plurality of keywords are input to the reception unit, a determination unit that determines the priority of the modification relationship by the plurality of keywords based on the input form or input order of the plurality of keywords;
An information search apparatus comprising: an output unit that searches and outputs information related to the plurality of keywords from a search target based on the priority. An input unit for inputting the keyword;
The information search device according to claim 3, further comprising a display unit that displays an output result of the output unit.   The information search apparatus according to claim 3, wherein the information to be searched includes image data. Computer,
A first step of accepting keyword input;
When a plurality of keywords are input in the first step, a second step of determining the priority of the modification relationship by the plurality of keywords based on the input form or input order of the plurality of keywords;
An information search program for causing a CPU to execute a third step of searching for and outputting information related to the plurality of keywords from a search target based on the determined priority.

Images