JP5663742B2 - Image search server and image information management method for image search server - Google Patents

Description

  The present invention relates to an image search server and an image management method for the search server. In particular, the present invention relates to an image search server that performs a search with a ranking given to search conditions and an image information management method of the image search server.

  With the spread of digital cameras and the like, it has become possible to generate image information by taking images easily. In addition, with the development of communication technology and computer technology, image information generated by shooting or the like is uploaded to a server or the like, and a large number of image information exists on the Internet. Further, by including the URL of the image information in the web page, it is easy to create a web page including the image information.

  In such a situation, a system is known in which a large number of image information existing on the Internet is stored in association with search information, and a search is performed according to a search condition (see, for example, Patent Document 1). A server of such a system stores a file name storing image information, a word included in a web page referring to the image information, and the like as search information. When a search condition such as a keyword is input from a client browser or the like, image information that matches the search condition is searched from a plurality of image information, and can be browsed by the browser or the like. At this time, the server ranks the searched image information in the order that matches the search condition. Thereby, it becomes possible to display image information in order of rank in a browser or the like.

Japanese Patent Application Laid-Open No. 2004-240992

  However, generally, image information has a larger data amount than text information and the number of image information that will newly exist on the Internet every day. The storage capacity required for storing increases.

As one embodiment of the present invention, an image information storage unit that stores image information acquired by first reading from an image providing apparatus connected by a communication network, and image information stored in the image information storage unit are provided. A log storage unit for storing a log transmitted to a terminal device via a communication network, and calculating a transmission frequency of the image information within a predetermined period from the log, and the transmission frequency exceeds a predetermined threshold, or A rate of increase in the transmission frequency of image information is calculated from the log, and when the rate of increase exceeds a predetermined threshold, a reread control unit that controls rereading of the image information from the image providing device; The corresponding image information is re-read from the image providing device under the control of the reading control unit, and the amount of information is increased before the first reading is stored in the image information storage unit. To provide an image search server and an image information reread unit re-stored in the image information storage unit.

As one embodiment of the present invention, image information acquired by first reading from an image providing apparatus connected by a communication network is stored in an image information storage unit, and the image information stored in the image information storage unit is stored in a terminal The log transmitted to the apparatus via the communication network is stored in the log storage unit, and the transmission frequency of the image information calculated by analyzing the log stored in the log storage unit within a predetermined period exceeds a predetermined threshold. Alternatively, when the rate of increase in the transmission frequency of image information calculated by analyzing the log stored in the log storage unit exceeds a predetermined threshold, the corresponding image information is reloaded from the image providing device. And providing an image information management method for an image search server including increasing the amount of information and storing the image information in the image information storage unit when the first reading is performed. .

  According to the present invention, it is possible to efficiently store image information to be searched by the image search server. As a result, the required storage capacity can be made smaller than in the prior art.

1 is a schematic configuration diagram of a system including an image search server according to an embodiment of the present invention. Functional block diagram of an image search server according to an embodiment of the present invention Example of search screen display Example of search result display An example of a table used to process a query for image information An example of a table to prevent duplicate reloading An example of a graph showing the relationship between image information and transmission frequency The flowchart of the process of the image search server which concerns on one Embodiment of this invention. Example of search result display An example of a table to prevent duplicate reloading Example of a table that stores logs Example of sliding window Functional block diagram of computer

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. In addition, this invention is not limited to the following forms, It can implement by performing various deformation | transformation.

  FIG. 1 is a schematic configuration diagram of a system 100 including an image search server according to an embodiment of the present invention. The system 100 includes an image search server 101, an image display device 102, and an image providing device 103. The image search server 101 and the image display device 102 can communicate via a communication network 104 such as the Internet. Further, the image search server 101 and the image providing apparatus 103 can communicate with each other via the communication network 104. In FIG. 1, only one image display device 102 and one image providing device 103 are shown. However, there are a plurality of image display devices 102 and image providing devices 103, and they are connected to the communication network 104. It may be. Similarly, a plurality of image search servers 101 may exist, and these may be connected to the communication network 104.

  The main operation of the system 100 is as follows. First, the image search server 101 reads and stores image information provided by the image providing device 103. The reading of the image information at this time is the first reading of the image information or the first reading of the image information. It can be assumed that the image information is not stored in the image search server 101 before the first reading and the first reading.

  At this time, the image search server 101 stores search information for searching the stored image information in association with each other. For example, the image providing device 103 is a web server that provides a web page and the like, acquires a web page stored in the image providing device 103 by a crawler or the like, and converts the word and image information included in the web page or the like. Link information (for example, URL (Uniform Resource Locator)) is acquired, and the URL to the word and the image information is transmitted to the image search server 101. The image search server 101 reads and stores image information from the image providing apparatus 103 using the transmitted URL, and stores the transmitted word in association with the search information.

  At this time, the image information may not be stored in the image search server 101 with the same amount of information as that stored in the image providing apparatus 103 but may be stored with a smaller amount of information. . For example, the information amount may be reduced and stored by reducing the size of the image represented by the image information or reducing the image quality (for example, lossy compression). This is because if all the image information is stored as it is, a large storage capacity is required. Further, the image providing apparatus 103 may not store the read image information itself, but only store link information to the image information. Even when link information is stored, the amount of information is usually smaller than the image information stored in the image providing apparatus 103.

  The image display device 102 is a terminal device on which a browser or the like operates. When an inquiry about image information stored in the image search server 101 is transmitted from a browser or the like and received by the image search server 101, the image search server 101 searches for image information according to the received inquiry. . When the search result is obtained, the search result is transmitted from the image search server 101 to the image display device 102. The search result can be expressed as a set of links to image information stored in the image search server 101. The image display apparatus 102 requests image information from the image search server 101 using the links included in the received link set, and the image search server 101 displays the requested image information in response to the request. To device 102.

  The image search server 101 stores a log of image information transmitted in response to a request from the image display device 102. The image search server 101 analyzes the log and detects image information having a high transmission frequency (transmission frequency). Then, image information that is frequently transmitted and satisfies a predetermined criterion is read again from the image providing apparatus 103 and stored. At this time, the image search server 101 stores a larger amount of information than when the image information is first stored. For example, the size is increased or stored in a higher quality state than when the image information is first stored. In addition, when only the link information to the image information is first stored, the image information itself is stored.

  The total number of image information that is frequently transmitted (the total number can be defined as, for example, the number of files when image information is stored in a file) is smaller than the total number of image information that is transmitted less frequently Is known empirically. Therefore, in this way, the information amount of image information that is transmitted less frequently is reduced and stored, and when image information that is transmitted more frequently is read and stored again, the information amount is increased and stored. Information can be stored efficiently. That is, the required storage area can be made smaller than when all image information is stored in the same state as that stored in the image providing apparatus 103.

  FIG. 2 shows a functional block diagram of the image search server 101. In FIG. 2, the image search server 200 includes an image information transmission unit 201, an image information storage unit 202, a log storage unit 203, a reread control unit 204, and an image information reread unit 205.

  The image information transmission unit 201 transmits the image information stored in the image information storage unit 202 to the image display device 102 via the communication network. As described above, prior to the transmission of the image information, an inquiry is received from the image display device 102, and in accordance with the inquiry, the image information stored in the image information storage unit 202 is searched, and the search result is displayed. The requested image information from the image display device 102 may be sent back to the image display device 102. Alternatively, without receiving an inquiry from the image display apparatus 102, a request for image information specified by a URL or the like may be received, and the requested image information may be transmitted.

  When an inquiry is received from the image display apparatus 102, the following processing may be performed. For example, when the image search server 200 receives a connection request from the image display device 102, the image information transmission unit 201 displays screen information in which an interface for search is described in HTML (Hyper Text Mark-up Language) or the like. 102. FIG. 3 illustrates a screen 301 in which display based on screen information is performed on the display of the image display apparatus 102. On the screen 301, an area 302 for inputting a search word as an inquiry for image information is displayed. In addition, a button 303 for transmitting an inquiry input in the area 302 to the image search server 200 is displayed.

  When the button 303 is pressed using a mouse or the like connected to the image display device 102, an inquiry input in the area 302 is transmitted to the image search server 200. When the image search server 200 receives the inquiry, the image information transmission unit 201 searches the image information stored in the image information storage unit 202. The search result is returned to the image display apparatus 102.

  FIG. 4 shows an example in which an image representing image information requested by the image display device 102 is displayed on the screen 301 of the display of the image display device 102 based on the search result returned to the image display device 102. In FIG. 4, images 401 to 410 are displayed on the screen 301. Images 401 to 410 are images represented by partial image information as a result of the search. A scroll bar 411 is displayed so that other images can be seen. The images 401-410 are displayed higher as the query is more relevant. Among the images displayed at the same height, the image displayed on the left is more relevant than the image displayed on the right. It can be made high.

  Note that the image display apparatus 102 does not need to request all of the image information included in the returned search results at once. For example, when a new image needs to be displayed on the screen as a result of the scrolling, the image search server may be requested for image information of an image that needs to be displayed. That is, in FIG. 4, ten images 401-410 are displayed. Therefore, it is sufficient to request 10 pieces of image information from among the image information included in the search result first. When the search result includes 11 or more pieces of image information, the image information other than the images 401-410 is displayed at a stage where the scroll bar 411 or the like is operated and an image other than the images 401-410 needs to be displayed. The image search server may be requested.

  FIG. 5 shows an example of a table used by the image information transmission unit 201 to process an inquiry about image information. In FIG. 5, a weight and an image ID are associated with each search term. As an example of the weight, there is a value indicating the degree of relevance of the image information indicated by the image ID with respect to the search word. For example, in the image information indicated by the image ID of image7055 for the word “Akiba” in FIG. 5, 100 is associated as a weight, 90 is associated with image9029, and 85 is associated with image04016. Yes. This indicates that the image information indicated by the image ID of image 7055 is more relevant to the word “Akiba” than the image information indicated by the image IDs of image 9029 and image04016. Therefore, in the display of the search result for the query “Akiba”, it is possible to display in the order of image 7055, image 9029, and image04016.

  In FIG. 5, the word “Akiba” is associated with a plurality of image IDs. In one embodiment of the present invention, the association between a word and an image ID is not limited to one-to-many, and may be many-to-one in which a plurality of words and one image ID are associated with each other. Further, many-to-many association may be used.

  Further, when a plurality of search terms are input as an inquiry, for example, a set of image IDs associated with each search term is acquired, and an image ID included in a common part of the set is calculated, It can be set as a search result. Further, the weight for the query can be defined as the minimum value of the weight for each search term. For example, for an image in which a weight of 100 is associated with “Akiba” and a weight of 80 is associated with “Present”, the weight for an inquiry made up of two search terms “Akiba Present” is , 80.

  The image information storage unit 202 stores image information. The stored image information is obtained by reading the image information stored in the image providing apparatus 103. As described above, the image information storage unit 202 detects that the web page provided by the image providing apparatus 103 is provided by a crawler or the like, and stores the read image information and link information thereof. Further, image information when image information having a high search frequency is re-read is also stored. Further, the image information to be reloaded may be image information that is frequently displayed as a search result or frequently clicked.

  The log storage unit 203 stores a log in which image information is transmitted. Here, “image information” is image information stored in the image information storage unit 202 and is image information transmitted to the image display apparatus 102. Further, a log of image information transmitted from another image search server to the image display apparatus 102 may be stored.

  The log storage unit 203 can also store a log of an inquiry received from the image display device 102 in order to search for image information. That is, the log storage unit 203 can also store an inquiry log received by the image information transmission unit 201. Further, the log storage unit 203 may store an inquiry log received by another image search server. Further, the image ID of the image information included in the search result obtained by the search by the image information transmitting unit 201 may be stored.

  Further, as a log stored in the log storage unit 203, which image is clicked when a search result is displayed on the screen 301 of the image display device 102 may be stored. In this case, when the search result is displayed on the screen 301 as shown in FIG. 4, the image information transmitting unit 201 determines which image is clicked when the image 401-410 is clicked. Is transmitted to the image search server 200, so that the search result is transmitted to the image display device 102.

  The reread control unit 204 controls the rereading of the image information based on the log stored in the log storage unit 203. That is, the log stored in the log storage unit 203 is analyzed, and transmitted image information having a high frequency is determined as a target for re-reading. For example, a threshold value is set, and image information that is transmitted more frequently within a predetermined period than the threshold value is to be reread. Further, image information that is displayed as a search result and frequently clicked may be determined as a target of re-reading. The image information determined as the reread target is read from the image providing apparatus 103 by the image information rereading unit 205 and stored in the image information storage unit 202.

  In addition, the re-reading control unit 204 does not compare the transmitted frequency or clicked frequency itself with a threshold value, but an image in which the increase rate of the transmitted frequency or clicked frequency exceeds another threshold value. Information may be reloaded. As a result, it is possible to re-read image information that is presumed to be rapidly attracting attention in the world.

  In addition, the reread control unit 204 may determine the image information to be reread by acquiring a log stored in a log storage unit such as another image search server or a web page search engine.

  Further, the re-reading control unit 204 calculates the weight of the search term for the image ID based on the log stored in the log storage unit 203 and stores it in the “weight” column of the table shown in FIG. The value may be updated. As a result of the update, the image information represented by the image ID whose weight satisfies a predetermined condition may be set as a reread target. For example, image information whose weight exceeds a predetermined threshold value is to be reread.

  In order to prevent duplicate rereading, the rereading control unit 204 may record which image information is the target of rereading. FIG. 6 shows an example of a table managed by the reread control unit 204 for this purpose. In the table shown in FIG. 6, information indicating whether or not re-reading has been performed and the URL of the image information are associated with the image ID. In FIG. 6, information indicating whether or not rereading has been performed is represented as a flag indicating values of NO and YES. NO indicates that rereading has not been performed. That is, as a result of the first detection by the crawler or the like, the read image information and link information are stored, and after that, no re-reading is performed. Moreover, YES indicates that the reread control unit 204 has been reread. When the re-read control unit 204 makes a re-read target, the image information is re-read according to the link information stored in the “original URL” column.

  For example, the image information whose value stored in the “reread” column of the table in FIG. 6 is YES is not subjected to rereading, and the rereading control unit 204 executes unnecessary rereading. Can be prevented. Also, by storing the time at which rereading is performed or the time at which rereading was actually performed in the table of FIG. 6 or the like, a predetermined time or more has elapsed even if rereading is performed once. As for image information, the latest image information can be acquired by re-reading the image information.

  The image information reread unit 205 reads the image information that has been read by the reread control unit 204 and stores it in the image information storage unit 202. For example, the image information is acquired from the image providing apparatus 103 according to the link information stored in the “original URL” of the table shown in FIG. Therefore, the image information corresponding to the image information read by the first reading (first reading) is reread and stored.

  Further, the image information that is to be read again and is read and stored in the image information storage unit 202 has a larger amount of information than when the image information is read and stored in the image information storage unit 202 by the first reading. Has increased. For example, when the amount of image information is reduced and stored in the image information storage unit 202 at the time of the first reading, the amount of information to be reduced is set as the target of re-reading, and the amount of information to be reduced is read first time. Make it smaller than the hour. Alternatively, the amount of information is not reduced.

  In the present embodiment, the image information stored in the image information storage unit 202 is re-read based on the result of log analysis by the log storage unit 203 by the re-read control unit 204. Accordingly, image information with a low transmission frequency can be stored with a small amount of information, and image information with a high transmission frequency can be stored with a large amount of information.

  FIG. 7 shows a graph obtained by arranging image information in descending order of image information transmission frequency and plotting the image information transmission frequency. That is, in FIG. 7, the transmission frequency of the image information N (for example, the image information with the Nth transmission frequency) is P. As shown in this graph, the total number of image information with a high transmission frequency is smaller than the total number of image information with a low transmission frequency. That is, most of the transmitted image information is occupied by a small number of image information. Therefore, for example, even if a transmission frequency threshold value is set, image information having a transmission frequency larger than the threshold value is stored in a large amount of information, and image information having a transmission frequency less than the threshold value is stored in a small information amount, The search result of the inquiry can include image information stored with a large amount of information. That is, it is possible to reduce the size of the storage capacity necessary for storing image information and to include image information with a large amount of information in the search results.

  FIG. 8 is a flowchart for explaining processing of the image search server 200 according to an embodiment of the present invention. As the processing in step S801, the log stored in the log storage unit 203 is read at predetermined time intervals. As the processing in step S802, the transmission frequency for each image ID is calculated as a result of reading the log. This transmission frequency may be a frequency transmitted within the predetermined time interval. In step S803, an image ID whose transmission frequency satisfies a predetermined condition is obtained. As the predetermined condition, for example, “transmission frequency exceeds a predetermined threshold” can be adopted. In step S804, the image of the URL whose “reread” flag is NO among the “original URL” associated with the obtained image ID is reread.

  In the embodiment of the present invention, the image information read by the image information reread unit 205 may be stored in the image information storage unit 202 as it is, or a process for changing to a predetermined size is added. Or a process for making the image clear may be added. For example, when the image is displayed on the screen 301, the image information re-read by the image information re-reading unit 205 is reduced in size so as to have a predetermined width so that the image information is displayed as an image having the same width. Alternatively, an enlargement process may be performed. As the predetermined width, 300 pixels or 250 pixels to 350 pixels can be adopted. By adopting this width, in many personal computers, the display width is around 1000 pixels, so that images can be displayed in a balanced manner in three columns. Further, as a process for clearing an image, a process for increasing the contrast or changing the overall brightness can be used.

  FIG. 9 shows a state in which the image information read by the image information re-reading unit 205 is displayed as a search result after the size is changed so as to have the same width. Images 901 to 907 are images having the same width. By adopting 300 pixels or 250 pixels or more and 350 pixels or less as the width as described above, the image can be displayed in a balanced manner in three columns.

  The number of columns may be changed without being fixed to 3. For example, when the window width in which an image is displayed is changed, the quotient value obtained by dividing the window width by the width of the image information can be used as the number of columns.

  FIG. 10 shows an example of another table for controlling the reloading. That is, another example of the table shown in FIG. When the table shown in FIG. 10 is compared with the table shown in FIG. 6, a column “maintain” is added to the table shown in FIG. 10. The value stored in this column indicates whether the image ID does not need to be read again. For example, if the image information is image information related to information that does not change with the passage of time, such as a dictionary, the image information is considered not to change, and therefore it is not necessary to re-read. Therefore, by storing “YES” as the value of the “maintain” column, the image information is not re-read. That is, the re-reading control unit avoids re-reading of image information in which “YES” is stored as the value of the “maintain” column. If the value in the “Maintain” column is “YES”, it is not necessary to manage whether or not the reread has been performed. Therefore, in FIG. 10, the value of the “reread” column is “−” in the row where the value of the “maintained” column is “YES”.

  Also, the number of image information to be re-read increases with the passage of time. For this reason, the size of the area for storing image information increases. Therefore, it is conceivable to change the image information whose transmission frequency has decreased to an image with a small information amount based on the result of log analysis by the log storage unit 203. Even in this case, the image information in which “YES” is stored as the value of the “maintain” column is not changed, so that unnecessary processing such as occurrence of re-reading can be prevented.

  Below, an example of the log memorize | stored in the log memory | storage part 203 and an example of the analysis of the log by the re-reading control part 204 are demonstrated. In this example, in addition to the log of transmitted image information, the log of inquiry is stored, the frequency that the search term as the inquiry is stored in the log is analyzed, and the image information that is predicted to increase rapidly is detected. doing.

  FIG. 11 shows an example of a log stored in the log storage unit 203. The log storage unit 203 stores the search term input (received) as an inquiry in association with the input time. For example, “Akiba” is input at 11:29:24 on June 22, 2010, “Freestyle” is input at 11:29:24 on June 22, 2010, and “Akiba” is It was entered again at 11:29:25 on June 22, 2010. Further, such a log of search terms indicates that there is a high possibility that image information associated with the recorded search terms is transmitted. For example, referring to an example of the association between the search term and the image ID in FIG. 5, “Akiba” input at 11:29:24 on June 22, 2010 has image 7055, image 9029, and image04016 as the image ID. This indicates that there is a high possibility that the image information that is included is transmitted. The possibility that the image information is transmitted increases as the weight increases. Therefore, image information having a weight equal to or greater than a predetermined value can be treated as being transmitted when analyzing a search term log.

  The re-reading control unit 204 divides the time into sections of a certain length, and associates the stored log with each section. As shown in FIG. 11, in the section (t−1), “Akiba” from 11:29:24 on June 22, 2010 to “Akiba” from 11:29:25 on June 22, 2010 In the section (t), a “present” from “Free Style” on June 22, 2010 at 11:29:27 to 11:29:30 on June 22, 2010 is associated.

  The re-reading control unit 204 stores the calculated reception frequency for each search term in the database for each section. FIG. 12 shows an example of this database. That is, the re-reading control unit 204 identifies the search word input for each section through the log data, counts the number of times received based on the identified search word, and stores it in the database.

  As shown in FIG. 12, the sliding window means a virtual time range that plays a role in determining information for searching the table shown in FIG. 12, and is displayed in n intervals set by the system operator. The information stored in the table shown in FIG. The value n is set to an arbitrary length as required by the system operator. In other words, the sliding window is a period between the present time and the time point that has been traced back by a predetermined time interval from the present time.

  The re-reading control unit 204 calculates time data belonging to n periods, which are sliding windows, and a search word / input count (reception count) corresponding to the time data from the table shown in FIG. It plays a role to add up. In other words, the re-reading control unit 204 calculates information on the search word input (received) to the image search server 101 and the cumulative number of times of input for each search word within the time range determined by the sliding window.

  The reread control unit 204 calculates the estimated number of inputs for each search term based on the total number of inputs. That is, the re-reading control unit 204 uses the number of times of input for each search term calculated within a limited time range close to the present time, and the number of times of input expected to occur in a certain period of time in the future. Play a role in predicting.

  In the calculation of the estimated number of times of input, the re-reading control unit 204 can predict the number of times of input using only the search word input for the search after removing the search word that has been artificially operated. . For example, when an advertising agency excessively inputs a specific search term with the intention of increasing the number of inputs, the re-reading control unit 204 determines that the search term is an aggressive search term, and the related input count Or a predetermined exclusion process is performed so that is not reflected in the calculation of the estimated number of inputs. In addition, the estimated number of inputs is calculated in consideration of seasonal fluctuation factors, which day of the week, time fluctuation factors such as morning or afternoon in a day, and the like.

  The reference time for estimating the estimated number of inputs is flexibly set by the system operator. For example, the reference time is estimated based on one day, or estimated based on one hour.

  FIG. 12 illustrates a sliding window that determines the time range of information used when predicting the estimated number of inputs. The sliding window is selected by time data “t to t−n” as n periods that are time ranges, and the re-reading control unit 204 selects a search word corresponding to the time data (t to t−n) and Identify the number of inputs. Thereafter, the re-reading control unit 204 adds the number of times of input for each search term using the identified information. The sliding window which determines the time range of the information used when predicting the frequency | count of input in the period of a fixed length in the future is illustrated. The sliding window is selected by time data “t to t−n” as an n period that is a time range, and the re-reading control unit 204 searches the search word and the number of inputs corresponding to the time data (t to t−n). Identify Thereafter, the re-reading control unit 204 adds the number of times of input for each search term using the identified information.

  The number of times of input for each search word belonging to the time data (t to t−n) can be accumulated as, for example, “Present -601 times, Freestyle -567 times, Akiba -487 times”. The re-read control unit 204 based on the number of times of input for each search term predicts the number of times of input for a certain period in the future for each search term, and for example, a conversion value that takes into account seasonal fluctuations, days of the week, etc. "127.4" selected as the number of input times multiplied by the number of times of each input, such as "Present-76567.4 times, Freestyle-7235.8 times, Akiba-61916.4 times" Can be predicted. The converted value is a numerical value that is flexibly set by the system operator, and the converted value “127.4” applied in the above example is optimally utilized by the operator of the present invention through sufficient experimentation and calculation. It is determined as a value determined to be possible.

  Therefore, for example, it can be predicted that the image information associated with the search word “present” will be transmitted about 76567 times during the next sliding window.

  The re-reading control unit 204 plays a role of calculating a determination numerical value using the calculated estimated number of inputs, and at the same time, the re-reading control unit 204 selects a search term whose input number increases rapidly based on the calculated determination numerical value. Play a role to detect. Here, the determination numerical value is numerical information for determining whether the related search word is a search word that rapidly increases, and the re-reading control unit 204 selects a search word whose determination numerical value is equal to or larger than a predetermined numerical value or relatively large. , It is determined and detected as a search term in which the number of inputs increases rapidly.

  As an example of the determination numerical value, the input rank for each search term is determined based on the total number of input times. When the input rank number becomes large (when the input rank is far from 1st place), the input rank number becomes small ( There is a first decision value that increases (when the input rank is close to 1). Such a first determination numerical value can be calculated as the reciprocal number of the number representing the rank (1 if the input rank is first and 3 if the input rank is third). In addition, there is a second determination numerical value calculated as the size of the estimated input frequency with respect to the minimum input frequency that is a predetermined frequency. For example, there is a value obtained by dividing the estimated number of inputs by the minimum number of inputs. Here, the “minimum number of inputs” means the minimum number of inputs for detection as a rapidly increasing search term.

  The re-reading control unit 204 has the second determination numerical value greater than or equal to a predetermined numerical value, and the first determination numerical value is higher than the first determination numerical value of the other search terms, and the input frequency rapidly increases. Detect as search term. Then, image information searched by such a search word is set as a reread target.

  Moreover, you may update the weight of the image information searched with such a search term, and a search term using either one or both of a 1st determination numerical value and a 2nd determination numerical value.

  In this way, by using two determination numerical values, it is possible to detect image information that is predicted to have a rapid increase in transmission frequency from the number of times a search word is input in one sliding window.

  The image search server according to an embodiment of the present invention can be realized by a computer. FIG. 13 shows a functional block diagram of the computer. The computer 1300 includes a CPU (Central Processing Unit) 1301, a RAM (Random Access Memory) 1302 as a main memory, a ROM (Read Only Memory) 1303 that stores programs to be executed after power-on, a secondary memory 1304, and the like. The communication I / F 1305 and the display I / F 1306 are connected to each other via a bus.

  When the computer 1300 is powered on, the program stored in the ROM 1303 is executed by the CPU 1301. By executing this program, a program constituting an operating system or the like stored in the secondary storage 1304 is loaded into the RAM 1302 and executed by the CPU 1301. As such a program, there is a program for realizing the operation of the image search server according to an embodiment of the present invention. Note that the operation of the image search server according to the embodiment of the present invention may be realized by a plurality of programs.

  For example, in order to realize the image information transmission unit 201, a request for image information is received by the communication I / F 1305, the requested image information is read from the secondary storage unit 1304, and transmitted by the communication I / F 1305. Further, for example, an inquiry is received by the communication I / F 1305, the image information is searched by the CPU 1301 by referring to the information in the table shown in FIG. 5 stored in the secondary storage 1304 or the RAM 1302, and the search result is displayed. It can also be transmitted by communication I / F 1305.

  Further, the log is stored in the secondary storage 1304 or the RAM 1302 in order to realize the log storage unit 203. In order to realize the image information storage unit 202, image information is also stored in the secondary storage 1304 and the RAM 1302. Further, in order to realize the re-read control unit 204, the log, the table structure shown in FIG. 5, FIG. 6 and FIG. 10 are read out from the information stored in the secondary storage 1304 and RAM 1302, and the above-described determination numerical values are obtained. Are calculated by the CPU 1301.

  In order to realize the image information re-reading unit 205, the image information to be re-read is transmitted by the communication I / F 1305, and the image information for which the acquisition request has been received is received and read to obtain the secondary information. It is stored in the storage 1304 and the RAM 1302.

  200 image search server, 201 image information transmission unit, 202 image information storage unit, 203 log storage unit, 204 reread control unit, 205 image information reread unit

Claims (16)

An image information storage unit for storing image information acquired by first reading from an image providing apparatus connected by a communication network;
A log storage unit for storing a log in which image information stored in the image information storage unit is transmitted to the terminal device via a communication network;
The transmission frequency of the image information within a predetermined period is calculated from the log, the transmission frequency exceeds a predetermined threshold, or the rate of increase in the transmission frequency of the image information is calculated from the log, and the rate of increase When the value exceeds a predetermined threshold, a re-reading control unit that controls re-reading of the image information from the image providing device;
Re-reading the corresponding image information from the image providing device under the control of the re-reading control unit, and increasing the amount of information than when stored in the image information storage unit at the time of the first reading An image search server comprising: an image information rereading unit for re-storing in an image information storage unit.   The image information acquired by the first reading from the image providing device and stored in the image information storage unit has a smaller amount of information than that stored in the image providing device. The image search server according to 1. Image search server according to claim 1 or 2, characterized in that the width of the image represented by the image information the image providing apparatus reload is from being stored in the image information storage unit has become a predetermined size. 4. The image search server according to claim 1, wherein the re-reading control unit records the re-read image information and excludes the recorded image information from a re-read target . 5. The image information is transmitted in response to an inquiry for searching for image information stored in the image information storage unit,
The image search server according to claim 1, wherein the log stored in the log storage unit includes a log of the inquiry. The image information storage unit stores image information in association with one or more keywords acquired from the image providing device;
The image search server according to claim 5, wherein the inquiry is an inquiry using a keyword. An image information storage unit for storing image information acquired by first reading from an image providing apparatus connected by a communication network;
A log storage unit for storing a log in which image information stored in the image information storage unit is transmitted to the terminal device via a communication network;
A reread control unit that calculates the transmission frequency of the image information from the log, and controls rereading of the image information from the image providing device;
Re-reading the corresponding image information from the image providing device under the control of the re-reading control unit, and increasing the amount of information than when stored in the image information storage unit at the time of the first reading An image information re-reading unit to be stored again in the image information storage unit;
Have
The image information is transmitted in response to an inquiry for searching for image information stored in the image information storage unit,
The log stored by the log storage unit includes the inquiry log,
The re-read control unit
Using the processor, calculate the number of receptions for each inquiry from the log stored in the log storage unit in a period between the present and the time point that is traced back in the past by a predetermined time interval from the present,
Using the processor, based on the calculated number of receptions, to calculate an estimated number of inputs predicted to occur in a future period of time for each inquiry,
Using the processor, the number of reception ranks for each inquiry is obtained based on the calculated number of receptions. The number of reception ranks increases, and the number of reception ranks increases, and the number of reception ranks increases. A first determination numerical value is obtained,
Using the processor, a second determination numerical value calculated as the magnitude of the estimated number of inputs with respect to a predetermined value is calculated for each inquiry.
Using the processor, the second determination numerical value is equal to or greater than a predetermined numerical value, and the calculated first determination numerical value is searched for inquiries relatively larger than the first determination numerical values of other inquiries. rows of planted image search server control to reload the image information reread unit.   The reread control unit holds an identification ID for identifying image information that does not need to be reread, and avoids the control of the reread control unit for rereading the image information identified by the identification ID. The image search server according to claim 1. Storing the image information acquired by the first reading from the image providing apparatus connected by the communication network in the image information storage unit;
Storing a log of the image information stored in the image information storage unit transmitted to the terminal device via a communication network in the log storage unit;
The transmission frequency of the image information calculated by analyzing the log stored in the log storage unit within a predetermined period exceeds a predetermined threshold, or is calculated by analyzing the log stored in the log storage unit When the rate of increase in the transmission frequency of the image information exceeds a predetermined threshold, the corresponding image information is reloaded from the image providing device and stored in the image information storage unit at the time of the first reading. An image information management method for an image search server, comprising: increasing the amount of information from time to time and storing the information in the image information storage unit again.   The image information acquired by the first reading from the image providing apparatus and stored in the image information storage unit has a smaller amount of information than when stored in the image providing apparatus. Item 10. The image information management method of the image search server according to Item 9. The image search server according to claim 9 or 10, characterized in that the width of the image represented by the image information the image providing apparatus reload is from being stored in the image information storage unit has become a predetermined size Image information management method. 12. The image information management method for an image search server according to claim 9, wherein the re-read image information is recorded, and the recorded image information is excluded from re-read targets. . The image information is transmitted in response to an inquiry for searching for image information stored in the image information storage unit,
The image information management method for an image search server according to claim 9, wherein the log stored in the log storage unit includes a log of the inquiry. When storing image information in the image information storage unit, the image information is stored in association with one or more keywords acquired from the image providing device,
The image information management method of the image search server according to claim 13, wherein the inquiry is an inquiry using a keyword. Storing the image information acquired by the first reading from the image providing apparatus connected by the communication network in the image information storage unit;
Storing a log of the image information stored in the image information storage unit transmitted to the terminal device via a communication network in the log storage unit;
Based on the transmission frequency of the image information calculated by analyzing the log stored in the log storage unit, the corresponding image information is re-read from the image providing device, and at the time of the first reading, Including increasing the amount of information than when stored in the image information storage unit and re-storing in the image information storage unit,
The image information is transmitted in response to an inquiry for searching for image information stored in the image information storage unit,
The log stored by the log storage unit includes the inquiry log,
Using the processor, calculate the number of receptions for each inquiry from the log stored in the log storage unit in a period between the present and the time point that is traced back in the past by a predetermined time interval from the present,
Using the processor, based on the calculated number of receptions, to calculate an estimated number of inputs predicted to occur in a future period of time for each inquiry,
Using the processor, the number of reception ranks for each inquiry is obtained based on the calculated number of receptions. The number of reception ranks increases, and the number of reception ranks increases, and the number of reception ranks increases. A first determination numerical value is obtained,
Using the processor, a second determination numerical value calculated as the magnitude of the estimated number of inputs with respect to a predetermined value is calculated for each inquiry.
Using the processor, the second determination numerical value is equal to or greater than a predetermined numerical value, and the calculated first determination numerical value is searched for inquiries relatively larger than the first determination numerical values of other inquiries. image information management method of the re-read row of planted image search server. The image of the image search server according to any one of claims 9 to 15, wherein an identification ID for identifying image information that does not need to be read is held, and re-reading of the image information identified by the identification ID is avoided. Information management method.

Images