JP2006302194A - Content output system and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To appropriately provide content which suits user's concerns according to the user's concerns. <P>SOLUTION: An information collection analyzer 70 acquires text data created and browsed by a personal computer 30 from the personal computer through a user interface 40 and acquires text data of voice transmitted and received with the other telephone system at the time of telephone conversation from a digital telephone system 50. Then, concerns of a user who uses the personal computer, etc. are presumed based on the data by every piece of acquired data. In addition, keywords are extracted from the acquired data by every piece of acquired data and contents corresponding to the keywords are acquired from a server. In addition, main concerns of the user at present are presumed based on transition of the presumed concerns of the user and the contents with high similarity to the main concerns are preferentially outputted. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for estimating a user's interest and providing the user with content corresponding to the estimated interest.

  2. Description of the Related Art A content providing system that searches content that is estimated to be of interest to a user based on a user input to a user interface and outputs the content to a user by monitoring the content is known. For example, there is known a content providing system that recognizes input sound from a microphone and outputs the content searched based on a keyword extracted from the sound (see Patent Document 1).

In this content providing system, content is acquired and played in order according to the extracted keywords. When the next content is not yet finished when trying to play the next content, the subsequent (next ) Hold content acquisition and playback for a certain period of time. If the previous content has been played back during that fixed time, after confirming that, the subsequent content is acquired and played back, but the previous content is played back during the fixed time. Is not completed, neither acquisition nor playback of subsequent contents is performed (see paragraph “0037” of Patent Document 1). In addition, when an attempt is made to play back the next content, if the previous content has not yet ended, the playback of the previous content is aborted, and the acquisition and playback of the subsequent (next) content is started. It is described in Document 1 (see paragraph “0053” of Patent Document 1).
JP 2004-164678 A

  However, the technique described in Patent Document 1 is useful for the user because when the content reproduction is not in time for the user's interests, the reproduction of the reserved content is stopped or the content being reproduced is terminated. Content may not be played or interrupted.

  The present invention has been made in view of these problems, and an object of the present invention is to provide a technique capable of providing a user with content that matches the user's interests in a more appropriate manner. To do.

  The content output system according to claim 1, which is made to achieve the above object, includes a data acquisition unit, a first interest estimation unit, a second interest estimation unit, a content acquisition unit, a content recording unit, Similarity evaluation means, output means, and content output control means are provided.

  The data acquisition means in the content output system is an information processing apparatus having a user interface, wherein at least one of data input to the user interface from the user and data output to the user by the user interface Get from.

  On the other hand, each time the data acquisition means acquires the data, the first interest estimation means estimates the user's interest based on the acquired data, and the second interest estimation means Is corrected based on the transition of the estimation results for the past predetermined period by the first interest estimation means, thereby estimating the current user's main interest.

  In addition, the content acquisition unit acquires content corresponding to each data acquired from the information processing device by the data acquisition unit from the device holding the content, and the content recording unit includes the content acquired by the content acquisition unit and the content And the estimation result of the first interest estimation means obtained from the acquired data from the information processing apparatus corresponding to is stored in the storage means in association with each other.

  In addition, the similarity evaluation unit is configured to estimate the result of the first interest estimation unit stored in the storage unit in association with each content stored in the storage unit, and the main user estimated by the second interest estimation unit. Evaluate similarity to interests.

  In addition, the output unit is configured to output each content stored in the storage unit to the user, and the content output control unit determines that each content stored in the storage unit is high in similarity by the similarity evaluation unit. Output from the evaluated content is preferentially output to the output means.

  In the content output system of the present invention configured as described above, the similarity with the main interest of the user estimated by the second interest estimation means is evaluated for each content of the output candidate, and the second interest estimation is performed. Since the means estimates the user's main interest by correcting the latest estimation result by the first interest estimation means, the evaluation value (similarity) of the content corresponding to the data recently acquired by the data acquisition means is Although it tends to be high, if the similarity with the user's main interest is high, the content is preferentially output to the user regardless of the acquisition time. Can be provided to the user with content that matches the user's interest appropriately.

  In the present invention, the second interest estimation means corrects the latest estimation result by the first interest estimation means and estimates the user's main interest. Since the user's interest is estimated based on the data, the fluctuation of the estimation result is large, and it is only necessary to evaluate the similarity with the latest estimation result by the first interest estimation means. This is because there is a possibility that content may not be output in an appropriate order because of excessive sensitivity to changes in interest.

  According to the present invention, the latest estimation result by the first interest estimation means is corrected based on the transition, the main interest of the user is estimated, the similarity with this is evaluated as described above, and the content Therefore, content that is highly useful to the user can be output appropriately.

As a method of estimating the interest, a method of extracting a keyword from the acquired data of the data acquisition means and estimating the user's interest based on the extracted keyword can be considered.
The content output system according to claim 2 includes an extraction unit that extracts a keyword from the acquired data every time the data acquisition unit acquires the data. The first interest estimation unit and the content acquisition unit include the following: It is comprised as follows.

  The first interest estimation means in this content output system estimates the user's interest each time the data acquisition means acquires the data based on the keyword extracted by the extraction means from the acquired data. In addition, the content acquisition unit acquires, for each data acquired by the data acquisition unit from the information processing device, content related to the keyword extracted by the extraction unit from the data from a device that holds the content (for example, a web server). To do.

  According to the content output system of claim 2 configured as described above, since the keyword is extracted and the user's interest is estimated, the user's interest is interpreted by interpreting the context of the sentence shown in the data. The interest can be estimated more easily than the case where the interest is estimated.

  Further, in the above-described invention, it has been described that a high similarity is preferentially handled when outputting content, but a content output system is configured so that a high similarity is preferentially handled when acquiring content. May be.

  The content output system according to claim 3 includes the data acquisition unit and the first interest estimation unit configured as described above, and the data recording unit, the second interest estimation unit, the content acquisition unit, the output unit, and the like described below. Degree evaluation means and content acquisition control means.

  The data recording unit in the content output system stores the estimation result of the first interest estimation unit and the acquired data from the information processing apparatus used to derive the estimation result in the storage unit in association with each other.

  Further, the second interest estimation means corrects the latest estimation result by the first interest estimation means based on the transition of the estimation result for the past predetermined period by the first interest estimation means, so that the current user's Estimate the main concern. In addition, the content acquisition unit acquires the content corresponding to each acquisition data from the information processing device stored in the storage unit from the device holding the content, and the output unit acquires each content acquired by the content acquisition unit, Output to the user.

  On the other hand, the similarity evaluation unit is configured to obtain an estimation result of the first interest estimation unit stored by the storage unit in association with each acquired data stored in the storage unit, and a main user estimated by the second interest estimation unit. The degree of similarity with the interest is evaluated, and the content acquisition control unit preferentially selects each acquisition data stored in the storage unit from the data evaluated as having a high similarity by the similarity evaluation unit. The content acquisition unit preferentially acquires the content corresponding to the selected acquisition data to be processed.

  In this content output system, as described above, content having a high degree of similarity with the user's main interest is prioritized from the device that holds the content (for example, a web server) regardless of the data acquisition time from the information processing device. And is output to the user.

  As described above, according to the present invention, in order to preferentially acquire highly useful content, the CPU load and the network load when acquiring the content are reduced, and the user's load is appropriately increased. Content that fits the interests can be acquired and provided to the user.

  The content output system can be applied with the same concept as that of the second aspect of the invention. According to a fourth aspect of the present invention, there is provided a content output system comprising an extracting unit that extracts a keyword from the acquired data every time the data acquiring unit acquires the data. The first interest estimation means in this content output system estimates the user's interest each time the data acquisition means acquires the data based on the keyword extracted by the extraction means from the acquired data.

  In addition, the data recording means associates the estimation result of the first interest estimation means with the keyword extracted by the extraction means, which is acquired data from the information processing apparatus used to derive the estimation result. To be stored in the storage means. In addition, the content acquisition unit acquires content related to the keyword extracted from each acquisition data stored in the storage unit from an apparatus that holds the content.

  According to this content output system, since the keyword is extracted and the user's interest is estimated as in the content output system according to claim 2, the user's interest is estimated. The interest can be estimated more easily than the case where the interest is estimated.

  In addition, in the content output system according to claims 2 and 4, specifically, when the first interest estimation means is configured as described in claim 5, the user's interest can be estimated easily and appropriately. This is convenient and the similarity evaluation is simplified.

  The first interest estimation means in the content output system according to claim 5, wherein each time the data acquisition means acquires the data, each keyword extracted by the extraction means from the acquired data is assigned to a plurality of predetermined classes. The class value for each class is obtained based on the keyword, and the user's interest is expressed as a vector based on the class value, and the user's interest is estimated.

  According to this content output system, the user's interest can be appropriately expressed by the type (class) of the keyword included in the acquired data from the information processing apparatus. Therefore, if the second interest estimation means represents the user's main interest in the same manner as the first interest estimation means, the similarity evaluation by the similarity evaluation means can be performed by taking the inner product of these vectors. Can be performed easily and appropriately. As a result, it is possible to provide the user with content that matches the user's interest more appropriately and matches the user's interest.

In addition, in the content output system according to claim 5 in which the first interest estimation unit is configured as described above, the second interest estimation unit may be configured as described in claim 6.
The second interest estimation means in the content output system according to claim 6 obtains a change amount of the estimation result expressed in the vector for the past predetermined period by the first interest estimation means, and uses the change amount to The latest estimation result by one interest estimation means is corrected, and the current main interest of the user is estimated.

  In the content output system configured as described above, the transition of the estimation result by the first interest estimation means can be appropriately evaluated by the temporal change of each element of the vector, and the current user's main concern can be determined. Can be estimated appropriately.

  More specifically, the second interest estimation means increases the weighting coefficient of the element whose value has increased and decreases the weighting coefficient of the element whose value has decreased in the estimation result of the first interest estimation means expressed in vector. And it can be set as the structure which correct | amends the newest estimation result by a 1st concern estimation means. If the second interest estimation means is configured in this way, it is possible to appropriately estimate the user's main interest.

Moreover, the function as each means in the content output system (Claims 1 to 6) described above can be realized by a computer by a program.
The program according to claim 7, in an information processing apparatus having a user interface, acquires from the information processing apparatus at least one of data input to the user interface from the user and data output to the user by the user interface. Each time the data acquisition means acquires the data, the first interest estimation means for estimating the user's interest based on the acquired data, and the latest estimation result by the first interest estimation means, The second interest estimation means for estimating the current user's main interest and the data acquisition means are acquired from the information processing device by correcting based on the transition of the estimation results for the past predetermined period by the first interest estimation means. Content acquisition that acquires content corresponding to each piece of data from the device that holds the content A content record that associates and stores the content acquired by the content acquisition means and the estimation result of the first interest estimation means obtained from the acquired data from the information processing device corresponding to the content in the storage means The degree of similarity between the estimation result of the first interest estimation means stored in the storage means and the main interest of the user estimated by the second interest estimation means for each content stored in the storage means The similarity evaluation means for evaluating the content, the output means for outputting the contents stored in the storage means to the user, and the contents stored in the storage means were evaluated as having high similarity by the similarity evaluation means. Storage means (for example, a semiconductor memory, etc.) functions as a content output control means for preferentially outputting the content to the output means. A computer having a program for implementing.

According to this program, an effect similar to that of the content output system according to claim 1 can be obtained.
The program according to claim 8 is an information processing apparatus including a user interface, wherein at least one of data input to the user interface from the user and data output from the user interface to the user is transmitted from the information processing apparatus. A data acquisition means to acquire, a first interest estimation means for estimating a user's interest each time the data acquisition means acquires data, an estimation result of the first interest estimation means, Corresponding data acquired from the information processing apparatus used for deriving this estimation result is stored in the storage means, and the latest estimation result by the first interest estimation means is the first interest By correcting based on the transition of the estimation result for the past predetermined period by the estimation means, the user's main concern at present Second interest estimation means for estimating, content acquisition means for acquiring content corresponding to each acquired data from the information processing apparatus stored in the storage means from the apparatus holding the content, and each content acquired by the content acquisition means For each acquisition data stored in the storage means and the estimation result of the first interest estimation means stored in the storage means in association with the output data and the second interest estimation means. The similarity evaluation means for evaluating the similarity with the user's main interest, and each of the acquired data stored in the storage means are preferentially processed from the data evaluated by the similarity evaluation means as having high similarity. Content acquisition control means for causing the content acquisition means to preferentially acquire content corresponding to the selected acquisition data to be processed A function as a storage means (e.g., a semiconductor manufactured memory, etc.) to a computer with a program for implementing.

  According to this program, an effect similar to that of the content output system according to claim 3 can be obtained.

  Embodiments to which the present invention is applied will be described below with reference to the drawings. The embodiments of the present invention are not limited to the following embodiments, and various forms can be adopted as long as they belong to the technical scope of the present invention.

[Description of configuration]
FIG. 1 is an explanatory diagram showing the configuration of a content providing system 1 to which the present invention is applied.
In the content providing system 1 of this embodiment, a web server 10 having various contents, an information search server device 20 and the like are connected to a WAN (wide area network) such as the Internet, and connected to the WAN via a router RT or the like. A personal computer (hereinafter referred to as “personal computer”) 30, a digital telephone device 50, and an information collection and analysis device 70 are connected to the LAN (Local Area Network).

(1) Information Search Server Device 20 The information search server device 20 includes a CPU, a ROM, a RAM, and the like (not shown), a control unit 21 that performs various arithmetic processes, and a storage unit 23 that includes a hard disk device and the like. And a communication unit 25 for communicating with other communication devices connected to the WAN, a database made up of link information (URL information) of various contents that the web server 10 has, various search programs, Is stored in the storage unit 23.

  The search program is under the web server software of the information search server device 20 and is executed by the control unit 21 when accessed from an external communication device. That is, when the information search server device 20 receives a query for the search program, the information search server device 20 searches the content corresponding to the search formula using the search program according to the search formula included in the query, and uses the link information of the content as the search result. (URL information) is returned to the sender of the query.

(2) About the personal computer 30 The personal computer 30 includes a CPU, a ROM, a RAM, and the like (not shown), performs a variety of arithmetic processing, a storage unit 33 including a hard disk device, a monitoring unit 35, and the like. , A communication unit 37 for communicating with other communication devices connected to the WAN or LAN, and a user interface 40 for transmitting user operation information to the control unit 31 and transmitting a calculation result of the control unit 31 to the user. And comprising. In addition, the personal computer 30 includes, as a user interface 40, an input unit 41 that can be operated by the user, such as a keyboard and a pointing device, and a display unit 43 that displays various types of information to the user.

  Software (application software) for a user to create a document through the input unit 41 or to view a document through the display unit 43 is installed in the storage unit 33, and the control unit 31 executes this software. As a result, a document file (data file) is created / updated according to the data input from the input unit 41, and text data (character information) included in the document file is displayed on the display unit 43. When a document file read command is input from the input unit 41, text data (character information) included in the document file is displayed on the display unit 43.

  In addition, the monitoring unit 35 monitors the software executed by the control unit 31, and the data that the software outputs to the display unit 43, and the text data (display target data) that the display unit 43 outputs to the user, The information is collected from the software and transmitted to the information collection and analysis device 70 through the communication unit 37. FIG. 2 is a flowchart showing personal computer monitoring processing executed by the monitoring unit 35 of the personal computer 30.

  When executing the personal computer monitoring process, the monitoring unit 35 determines in S110 whether there is software that holds uncollected display target data among the predetermined data collection target software. If it is determined that there is no software that holds the collected display target data (No in S110), it waits until uncollected display target data is generated, and if it is determined that there is software that holds the uncollected display target data ( The timer is reset (S120).

  When the processing in S120 is completed, the monitoring unit 35 determines whether the software having the uncollected display target data is in the document creation mode or the document browsing mode (S130). If it is determined that the mode is the creation mode (Yes in S130), device usage information indicating “document creation” is generated (S140), and if it is determined that the mode is not the document creation mode (that is, the document viewing mode) (S130). No), device usage information meaning “document browsing” is generated (S145).

  The document creation mode referred to here indicates that the software is updating the document file and display contents sequentially based on the data input by the user, and the document viewing mode refers to the user. The text data (character information) included in the document file designated by the user is displayed. In the present embodiment, it is assumed that the data collection target software is configured to notify the monitoring unit 35 of any one of the document creation mode and the document browsing mode.

  When the device usage information is generated in S140 or S145, the monitoring unit 35 is text data that the software outputs to the display unit 43 as display target data, and the uncollected text data from the software is displayed in the software. And is written in the transmission target file (S150). At this time, the monitoring unit 35 sequentially displays uncollected text data that is display target data output to the display unit 43 by the control unit 31 during a unit time T1 (for example, 1 minute). By writing to the transmission target file, the text data for the unit time T1 is stored in the transmission target file.

  Then, when the unit time T1 has passed (Yes in S160), source generation time information indicating the current time is generated (S170), and then the transmission target file having the generated text data for the unit time T1, and S170 The source generation time information generated in step S3, the input source information indicating “personal computer”, and the device usage information generated in S140 or S145 are associated with each other and transmitted to the information collection analysis device 70 (S180).

  When the processing in S180 is completed, the monitoring unit 35 determines whether or not an end command is input from the control unit 31 or the like (S190), and determines that no end command is input (No in S190). The process proceeds to S110 and waits until the next uncollected data is generated. If it is determined that an end command has been input (Yes in S190), the personal computer monitoring process is terminated.

(3) Digital Telephone Device 50 Returning to FIG. 1, the digital telephone device 50 is provided with a CPU, ROM, RAM, etc. (not shown), and stores a control unit 51 that performs various arithmetic processes, setting information, and the like. Storage unit 53, monitoring unit 55, communication unit 57 for communicating with other communication devices connected to WAN or LAN, and transmitting user input to internal devices (control unit 51 and communication unit 57) And a user interface 60 for transmitting the output of the internal device (the control unit 51 and the communication unit 57) to the user.

  The communication unit 57 is configured to be able to communicate with an external telephone device via the WAN (for example, a configuration having an IP (Internet Protocol) telephone function). The digital telephone device 50 is used as a user interface 60 by a user. A sound input unit 61 composed of a microphone or the like that collects the voice produced by converting the voice to sound data; a sound output unit 63 composed of a speaker or the like that reproduces sound data received by the communication unit 57 from an external telephone device; Is provided.

  In addition, the monitoring unit 55 acquires audio data input from the sound input unit 61 to the communication unit 57 via the control unit 51 and audio data input from the communication unit 57 to the sound output unit 63 via the control unit 51. These are converted into text data for each voice data of unit time T2 minutes (for example, 10 seconds), a transmission target file is generated, and this is transmitted to the information collecting and analyzing apparatus 70. . FIG. 3 is a flowchart showing a telephone monitoring process repeatedly executed by the monitoring unit 55 of the digital telephone device 50.

  When the telephone monitoring process is executed, the monitoring unit 55 stands by until the connection between the digital telephone device 50 and the other telephone device is made, and when the connection between the digital telephone device 50 and the other telephone device is made ( In S210, it is determined whether or not the connection is caused by a call from the digital telephone device 50 (S220), and it is determined that the connection is caused by a call from the digital telephone device 50. Then, device usage information meaning “call” is generated (S230), and when it is determined that the connection is caused by a call from another telephone device (incoming call from the digital telephone device 50), Device usage information indicating “incoming call” is generated (S235). Thereafter, the process proceeds to S240.

  If transfering it to S240, the monitoring part 55 will acquire the audio | voice data for unit time T2, convert the acquired audio | voice data into text data, and produce | generate the transmission object file which stored this. Further, source generation time information indicating the current time is generated (S250), and then the generated transmission target file, the source generation time information generated in S250, and input source information meaning “digital telephone device” The device usage information generated in S230 or S235 is associated with each other and transmitted to the information collection and analysis device 70 (S260).

  When the process in S260 is completed, the monitoring unit 55 determines whether or not the connection has been disconnected. When the monitoring unit 55 determines that the connection has not been disconnected (No in S270), the process proceeds to S240 for the next time. The voice data for T2 is converted into text data, which is transmitted to the information collecting and analyzing apparatus 70. If it is determined that the connection has been disconnected (Yes in S270), the telephone monitoring process is temporarily terminated. And it waits until the next connection is made (S210).

(4) Information Collection and Analysis Device 70 Returning to FIG. 1, the information collection and analysis device 70 constituting the content providing system 1 of this embodiment includes a CPU, a ROM, a RAM, and the like (not shown), and performs various operations. Content acquired from a control unit 71 that performs processing, a storage unit 73 that includes a hard disk device, a communication unit 75 that communicates with other communication devices connected to a WAN or a LAN, and the web server 10 Is displayed to the user, and when the content includes audio data, a sound output unit 79 that reproduces and outputs the audio data is provided.

  The information collection and analysis device 70 includes various programs executed by the control unit 71 and various tables used when executing the programs in the storage unit 73, and queries the information search server device 20 by executing the various programs. Information (for example, URL) that specifies the content of the web server 10 on the WAN that corresponds to the search key set in the query is acquired from the information search server device 20 as a search result. Further, the web server 10 is accessed based on the search result, content provided by the web server 10 is acquired, and displayed / reproduced.

  FIG. 4 is an explanatory diagram showing the configuration of the content providing screen displayed by the display unit 77. The content providing screen is an area for displaying text and images, and a text / image area configured to scroll and display new information, a video playback area for displaying a playback result of the video file, A term display area for displaying the term search result.

  The information collection and analysis device 70 issues a query based on the text data included in the file (the transmission target file) received from the personal computer 30 or the digital telephone device 50, and displays the content to be displayed on the content providing screen as a web server. Get from 10. These are provided to the user in the form of display / sound output.

In the following, various processes executed by the control unit 71 in order to realize this content providing service, and a configuration of a table used when the processes are executed will be described.
(4-1) Source Information Registration Process FIG. 5 is a flowchart showing the source information registration process repeatedly executed by the control unit 71.

  When the source information registration process is executed, the control unit 71 receives data transmitted from the personal computer 30 in S180 (see FIG. 2) or data transmitted from the digital telephone device 50 in S260 (see FIG. 3). Is received (S310), and when received (Yes in S310), the file (the transmission target file) included in the received data is stored in the storage unit 73 (S320). Furthermore, source information including source generation time information, input source information, device usage information, and the name of the file received together with the file is generated and registered in the source management table (S330). Thereafter, the source information registration process is temporarily ended, and the process waits until the next data is received (S310).

  As shown in FIG. 6, the source management table is configured to store a plurality of source information including source generation time information, input source information, device usage information, and file names. Based on the source information registered in the source management table, the received file (the transmission target file) is read from the storage unit 73 and the contents of the received file are analyzed.

(4-2) Keyword extraction processing (including interest vector calculation processing)
FIG. 7 is a flowchart showing a keyword extraction process executed by the control unit 71 in parallel with the source information registration process.

  When the keyword extraction process is executed, the control unit 71 determines whether or not unprocessed source information for the process of S415 exists in the source management table (S410). If it is determined that the unprocessed source information does not exist in the source management table (No in S410), the process waits until the unprocessed source information is generated by the source information registration process. If it is determined that the file exists in the table (Yes in S410), the text data of the file associated with the unprocessed source information is analyzed (morphological analysis), nouns in the text data are extracted, and a predetermined identification is performed. A noun other than the noun is extracted as a keyword (S415). In this embodiment, it is assumed that a general noun that does not function as a keyword when searching for content is stored in the storage unit 73 as a specific noun.

  When the process in S415 is completed, the control unit 71 proceeds to S420 and executes an interest vector calculation process. FIG. 8 is a flowchart showing an interest vector calculation process executed by the control unit 71. In the present embodiment, the user's interests are expressed in vectors, and this is used to determine the output priority of the content provided to the user. Although details will be described as appropriate as necessary, the “interest vector” in the present embodiment is a vector obtained by quantifying the user's interest. Specifically, the interest vector V is expressed as a 1-row n-column vector having a corresponding element for each predetermined type of interest.

V = (v1, v2,..., Vn)
When the interest vector calculation process is executed, the control unit 71 first resets the interest vector V to zero (S510), and from the keyword group extracted in S415 (see FIG. 7), the processes after S530 are performed. One unprocessed keyword is selected as a processing target (S520), and the selected processing target keyword is searched in the interest estimation table stored in the storage unit 73 (S530). FIG. 9 is an explanatory diagram showing the configuration of the interest estimation table.

  The interest estimation table includes, for each predetermined interest type, keywords belonging to the interest type and definition information (weighting definition information) regarding weighting performed when calculating the interest vector.

  Returning to FIG. 8, when the keyword to be processed is present in the keywords registered in the interest estimation table, the control unit 71 determines Yes in S540 and based on the interest estimation table. The type of interest to which the keyword to be processed belongs is determined (S550). Then, 1 is added to the value of the element vi of the interest vector V corresponding to the type of interest i (i = 1, 2,..., N) to which the keyword to be processed belongs (S560). Thereafter, the process proceeds to S570. On the other hand, if the keyword to be processed is not a keyword registered in the interest estimation table, the control unit 71 determines No (no search target) in S540, and proceeds to S570.

  When the process proceeds to S570, the control unit 71 determines whether or not an unprocessed keyword that has not been processed after S530 exists in the keyword group extracted in S530, and if an unprocessed keyword exists. If it judges (Yes in S570), it will transfer to S520 and will choose one keyword from among unprocessed keywords as a candidate for processing, and will perform processing after S530.

  In the present embodiment, when there are a plurality of the same keywords in the keyword group extracted in S530, even if the processing after S530 is executed for one keyword, other keywords (the same keyword) ) Is not processed, and the processing after S530 is executed for each keyword (each of the same keyword).

  If it is determined that there is no unprocessed keyword (No in S570), the value of each element vi of the interest vector V is weighted according to the weighting definition information in the interest estimation table (S580).

  The weighting definition information in the interest estimation table is classified into two types of definition information, and the first type of weighting definition information includes a combination of device usage information and a weighting coefficient W. On the other hand, the second type of weighting definition information includes a combination of time information and a weighting coefficient W.

  The control unit 71 uses the source information corresponding to the file in which the keyword is extracted in S415 (see FIG. 7) for the element vi of the interest vector corresponding to the type of interest for which the first type weighting definition information is set. Weighting is performed according to the weighting definition information only when the information matches the device usage information indicated by the weighting definition information.

  That is, in S580, the control unit 71 corresponds to the type of interest in which the first type of weighting definition information that matches the device usage information of the source information corresponding to the file from which the keyword is extracted in S415 (see FIG. 7) is set. An element vi of the interest vector to be weighted is set as a weighting target, and the weighting target element vi is multiplied by a weighting factor W.

vi ← vi × W
In addition, the control unit 71, for the interest vector element vi corresponding to the type of interest for which the second type of weighting definition information is set, the current time is within the range indicated by the time information of the weighting definition information Only according to the weighting definition information, weighting is performed.

  That is, in S580, the control unit 71 sets the interest vector element vi corresponding to the type of interest for which the second type of weighting definition information having the current time within the range indicated by the time information is set as a weighting target. The element vi to be weighted is set to a weighting factor W times.

When the weighting for each element ends in S580, the control unit 71 completes the calculation of the interest vector V, outputs this as a return value, and ends the interest vector calculation process.
Returning to FIG. 7, when the interest vector calculation process described above is completed in S420, the control unit 71 proceeds to S425, and in the interest vector V, the element that takes the maximum value (that is, the value of the value is higher than other elements). The type of interest corresponding to (large element) is determined (S425).

  When the processing in S425 is completed, a new writing area for the individual search information is secured in the search information management table stored in the storage unit 73 (S430), and the source information included in the source information from which the keyword is extracted in S415 is stored in that area. The generation time information is written (S435), the interest vector calculated in S420 is written (S440), the keyword extracted in S415 is written as a search key (S445), and the number of the search keys is written (S450).

  FIG. 10 is an explanatory diagram showing the configuration of the search information management table. As shown in FIG. 10, each individual search information includes source generation time information, an interest vector value, search key information, and search key number information.

  That is, the control unit 71 registers the individual search information corresponding to the source information extracted in S415 in the search information management table by executing the processes of S430 to S450. When the process of S450 is completed, the process proceeds to S455, where it is determined whether an end command is input. When it is determined that the end command is not input (No in S455), the process proceeds to S410, The process waits until unprocessed source information is generated. If it is determined that an end command has been input (Yes in S455), the keyword extraction process ends.

(4-3) Search Control Processing Next, search control processing that the control unit 71 executes in parallel with the keyword extraction processing will be described. FIG. 11 is a flowchart showing search control processing executed by the control unit 71.

  When the search control process is executed, first in S610, it is determined whether or not unprocessed individual search information exists in the search information management table for the process of S670, and if it is determined that there is unprocessed individual search information ( If it is determined that there is no unprocessed individual search information (No in S610), the process waits until unprocessed individual search information is generated.

  In S630, the control unit 71 selects the individual search information that is unprocessed individual search information and has the latest source generation time as the processing target. If there are a plurality of pieces of individual search information to be processed in S630, the individual search information having the largest number of search keys is selected again as a processing target (S650).

  When the processing in S650 is completed, the control unit 71 selects one piece of search program information that has not been issued from the search program management table (S660). Then, the individual search information selected in S650 is designated as a processing target, and a search process for the selected search program information is started (S670).

  FIG. 12 is an explanatory diagram showing the configuration of the search program management table stored in the storage unit 73. As shown in FIG. 12, the search program management table includes, for each search program to be executed, the address information (URL information) of the information search server device 20 having the search program, and the query generation method (search formula generation method). Search program information including query generation information indicating that the search program includes the category information indicating the type of content searched by the search program.

  That is, in S660, the control unit 71 searches the search program information included in the search program management table, and does not transmit the query to the URL-destination information search server device 20 according to the address information (search that has not been issued). Program information) is selected as a processing target, and a search process is started in S670. The activated search process is managed by the scheduler and executed in parallel with other processes, and generates a query based on the query generation information indicated by the specified search program information. The URL information indicated by the address information is transmitted to the information search server device 20.

  When the processing in S670 is completed, the control unit 71 determines whether search program information that has not been issued is present in the search program management table (S680), and determines that search program information that has not been issued is present. Then (Yes in S680), the process proceeds to S660, where one search program information to be processed is selected from the search program information that has not been issued a query, and the processes after S670 are executed.

  On the other hand, if it is determined that the search program information that has not been issued is not present in the search program management table (No in S680), it is determined whether an end command is input (S690), and no end command is input. (No in S690), the process proceeds to S610. On the other hand, if it is determined that an end command has been input (Yes in S690), the search control process ends.

(4-4) Search Processing Next, search processing realized in each search process executed in parallel by the control unit 71 will be described. FIG. 13 is a flowchart showing search processing.

  When the search process is executed, the control unit 71 first reads the selected search program information from the search program management table (S705), and generates the query generation information included in the read search program information and the individual processing target specified. This is an inquiry signal to the information search server device 20 that generates a search expression in which the search key of the individual search information is set as a search key based on a group of search keys of the search information and stores the search expression. A query is generated (S710).

  Then, the generated query is transmitted to the URL destination indicated by the search program information, and the information search server device 20 is caused to execute the URL destination search program of the information search server device 20 (S715). At this time, the information search server device 20 searches the content corresponding to the search key included in the individual search information according to the search formula included in the received query, and stores the link information (URL information) of the content as the search result. The response signal is returned to the information collection and analysis device 70.

  After transmitting the query in S715, the control unit 71 determines whether or not a response signal has been received from the information search server device 20 (S720), and determines that it has not received a response signal (No in S720). Wait until a response signal is received.

  On the other hand, if it is determined in S720 that a response signal has been received (Yes in S720), the process proceeds to S725 to determine whether the response signal is a search result or a response signal indicating that the search has been interrupted. If the response signal is not a search result (ie, a response signal indicating that the search is interrupted) (No in S725), the search process (search process) is terminated.

  On the other hand, when determining that the response signal is a search result (Yes in S725), the control unit 71 determines whether or not the category information of the specified search program information is “term search” (S730), “ If it is determined that the search is not “term search” but “general search” (No in S730), the search result consisting of a group of link information received as a response signal is used for a predetermined number of cases (for example, 5 searches) for general search. Link information (URL information) is extracted, individual result information describing the corresponding link information is generated for each link information (S735), and the storage unit 73 stores the individual result information for each generated link information. Is registered in the search result management table (S740).

  The individual result information includes source generation time information, category information indicating the type of search program that has output the search result, content storage destination address information indicating link information (URL information), and the cause of obtaining link information. The storage location information of the individual search information, the acquisition presence / absence information, the acquisition completion time, the output presence / absence information, and the evaluation value.

FIG. 14 is an explanatory diagram showing the configuration of the search result management table stored in the storage unit 73. The search result management table includes one or a plurality of individual result information.
In S735, the control unit 71, for each extracted link information, the source generation time information of the individual search information to be processed, the category information, the content storage destination address information describing the link information, and the processing The storage destination information of the target individual search information, the acquisition presence / absence information set to “NO”, the acquisition completion time information set to “NULL”, the output presence / absence information set to “NO”, and “NULL” The individual result information including the evaluation value set to "" is generated, and each individual result information is registered in the search result management table (S740). When the process in S740 is finished, the search process (search process) is ended.

  On the other hand, if it is determined in S730 that the category information of the designated search program information is “term search” (Yes in S730), the control unit 71 obtains a search result including a group of link information received as a response signal. From this, link information (URL information) for a predetermined number (for example, one) for term search is extracted, and individual result information (similar to that described above) describing the link information is generated. (S745) The generated individual result information is registered in the search result management table stored in the storage unit 73 (S750). Then, when the process in S750 is completed, the search process (search process) is ended.

(4-5) Search result evaluation process (including read process and current interest table update process)
Next, a search result evaluation process executed by the control unit 71 in parallel with the search control process will be described. FIG. 15 is a flowchart showing search result evaluation processing executed by the control unit 71.

  When the search result evaluation process is executed, the control unit 71 first determines whether or not unprocessed individual result information exists in the search result management table for the processes after S820 (S810). If it is determined that the individual result information exists (Yes in S810), the process proceeds to S820, and if it is determined that there is no unprocessed individual result information (No in S810), the unprocessed individual result information is stored in the search result management table. Wait until you are registered.

  When the process proceeds to S820, the control unit 71 selects unprocessed individual result information as a processing target, and when the process in S820 is completed, the process proceeds to S830 and executes a reading process. FIG. 16 is a flowchart showing a reading process executed by the control unit 71.

  When the reading process is executed, the control unit 71 first resets the timer (S905). Thereafter, the content storage destination address information included in the designated individual result information to be processed is read from the search result management table (S710), and the content request signal is transmitted to the URL destination indicated by the read content storage destination address information. (S915).

  After transmitting the request signal in S915, the control unit 71 determines whether or not a predetermined processing time limit has elapsed based on the timer value (S920), and if the processing time limit has not elapsed. If it judges (No in S920), it will transfer to S930. On the other hand, if it is determined that the processing time limit has elapsed (Yes in S920), the session is closed (communication is terminated) in S925, and the reading process (reading process) is ended.

  When the process proceeds to S930, the control unit 71 determines whether or not reception of the response signal for the request signal transmitted in S915 is completed, and determines that reception of the response signal is not completed (No in S930), S920. Migrate to On the other hand, if it is determined in S930 that reception of the response signal is completed (Yes in S930), the control unit 71 determines whether the response signal is requested content or error information (S940), and the response. If it is determined that the signal is not content (error information) (No in S940), the reading process (reading process) is terminated.

  On the other hand, if it is determined that the response signal is content (Yes in S940), the content is stored in the storage unit 73 (S945), and then the content storage destination address information included in the individual result information to be processed is converted to the content. The search result management table (individual result information) is updated by changing the URL information to the storage destination (S950). Thereafter, the reading process (reading process) is terminated.

  Returning to FIG. 15, when the reading process is completed, the control unit 71 proceeds to S840 and determines whether or not the content can be normally read in the reading process in S830. This determination is made based on whether or not the reading process has been completed through S950 (see FIG. 16) (if the reading process has been completed through S950, it is determined that the content has been successfully read out, and the process has not been performed through S950). When the reading process is completed, it is determined that the content could not be read normally).

  If it is determined in S840 that the content has been successfully read in the reading process in S830 (Yes in S840), the process proceeds to S845. On the other hand, if it is determined that the content could not be read normally in the reading process in S830 (No in S840), it is determined whether an end command is input (S880), and it is determined that no end command is input. Then (No in S880), the process proceeds to S810. On the other hand, if it is determined that an end command has been input (Yes in S880), the search result evaluation process ends.

  In S845, the control unit 71 updates the acquisition result information of the individual result information corresponding to the content that has been successfully read in the reading process in S830 to “YES” and enters the acquisition completion time information area. Write the current time. Thereafter, the process proceeds to S850, and the current interest table update process is executed. FIG. 17 is a flowchart showing current interest table update processing executed by the control unit 71.

  When the current interest table update process is executed, the control unit 71 reads the interest vector of the individual search information with the latest source generation time in the search information management table (S961). In the period in which the source information is generated one after another, the latest interest vector of the different individual search information is read each time in the process of S961, but in the period in which the generation of the source information is interrupted. In the process of S961, the interest vector of the individual search information corresponding to the source information generated immediately before the interruption is repeatedly read out.

  When the process in S961 is completed, the control unit 71 determines whether or not history information (information including registration time information and a cumulative interest vector) is registered in the interest history table included in the storage unit 73 (S963). . FIG. 18A is an explanatory diagram showing the configuration of an interest history table that the storage unit 73 has. The interest history table is configured to store history information up to a predetermined maximum number (three in this embodiment).

  Specifically, in S963, if one piece of history information is not registered in the interest history table, it is determined No, and the process proceeds to S965. If even one piece of history information is registered in the interest history table, Yes. The process proceeds to S971.

  After shifting to S965, the control unit 71 uses the current time as the registration time, generates the history information using the interest vector of the individual search information with the latest source generation time read out in S961 as the cumulative interest vector, and uses this as the interest history table. Register with. Thereafter, the current time is set as the update time, and the interest vector of the individual search information read in S961 is registered as the current interest vector Vc in the current interest table (S977). Thereafter, the current interest table update process is terminated.

  FIG. 18B is an explanatory diagram showing the configuration of the current interest table that the storage unit 73 has. The current interest table is a table composed of only one record, and the update time information indicating the update time of the record is associated with the current interest vector Vc handled as an interest vector representing the current user's main interest. Remember.

  Further, when the determination in S963 is Yes and the process proceeds to S971, the control unit 71 reads the individual search read in S961 into the accumulated interest vector in which the registration time registered in the interest history table is recorded in the latest history information. A cumulative interest vector is newly generated by adding the interest vectors of the information.

  If the history information registered in the interest history table is less than the maximum number (three), it is determined Yes in S973, and this newly generated cumulative interest vector is registered in the interest history table ( S975). Specifically, history information including the accumulated interest vector generated in S971 and registration time information describing the current time is generated and registered in the interest history table (S975). When this process (S975) is completed, the current time is set as the update time, the interest vector of the individual search information read in S961 is registered as the current interest vector Vc in the current interest table (S977), and then Then, the current interest table update process is terminated.

  On the other hand, if it is determined that the history information registered in the interest history table has the maximum value (three) (No in S973), the history with the oldest registration time is selected from the history information registered in the interest history table. Information is selected as a deletion target, and this is deleted from the interest history table (S981). Further, history information including the accumulated interest vector generated in S971 and the registered time information describing the current time is generated, and this is registered in the interest history table (S983).

Thereafter, the history information in the interest history table is sorted (rearranged) in the order of registration time, and the interest history table is updated (S985).
When the process in S985 is completed, the process proceeds to S987, and a transition coefficient K = (k1, k2,..., Kn) is calculated. Specifically, the transition coefficient K is obtained according to the following equation as shown in FIG.

K = ((v11−v12) / (v12−v13), (v21−v22) / (v22−v23),..., (Vn1−vn2) / (vn2−vn3))
However, the accumulated interest vector of the history information with the newest registration time among the history information registered in the interest history table is expressed as (v11, v21,..., Vn1). Further, among the history information registered in the interest history table, the accumulated interest vector of the history information having the second latest registration time is expressed as (v12, v22,..., Vn2). Then, among the history information registered in the interest history table, the cumulative interest vector of the history information having the third newest registration time (in other words, the oldest registration time) is represented by (v13, v23,..., Vn3). Express.

  In this way, after calculating the transition coefficient K in S987, the control unit 71 proceeds to S989 and calculates the interest vector V = (v1, v2,..., Vn) of the individual search information read in S961. The current interest vector Vc is calculated by multiplying the transition coefficient K = (k1, k2,..., Kn) for each element.

Vc = (k1, v1, k2, v2,..., Kn, vn)
The current interest vector Vc is registered in the current interest table (S990). In this case, the current interest vector Vc is registered with the current time as the update time. When the current interest vector in the current interest table is updated in this way, the control unit 71 ends the current interest table update process.

  Returning to FIG. 15, when the current interest table update process is completed, the control unit 71 proceeds to S860 and is associated with the individual result information in the storage destination information based on the individual result information selected in S820. The inner product of the interest vector V of the individual search information and the interest vector Vc of the current interest table is calculated, and this is written as an evaluation value in the evaluation value area of the individual result information selected in S820 in the search result management table. This inner product calculation is performed as follows.

  First, a normal vector E (also referred to as a unit vector) for the interest vector V is calculated by dividing the interest vector V of the individual search information by the norm || V || of the interest vector V.

E = V / || V ||
Further, the normal vector Ec for the current interest vector Vc is calculated by dividing the current interest vector Vc by the norm || Vc || of the current interest vector V.

Ec = Vc / || Vc ||
Thereafter, the inner product R of the normal vector E for the interest vector V and the normal vector Ec for the current interest vector Vc is calculated.

R = E ・ Ec
As a result, the inner product is calculated, and the individual result information to be processed in the search result management table is updated.

  When the process in S860 is completed, the control unit 71 proceeds to S870, and again sets the evaluation value area of the individual result information to 0 only for the evaluation value written in S860 that is less than the threshold value. Write. The threshold mentioned here should be determined by experiment, user preference, and the like, and by adjusting this threshold, it is possible to adjust the determination of whether or not content is output.

  When the process in S870 is completed, it is determined whether or not an end command is input (S880). If it is determined that no end command is input (No in S880), the process proceeds to S810. On the other hand, if it is determined that an end command has been input (Yes in S880), the search result evaluation process ends.

(4-6) Output processing (including display control processing)
Next, output processing that the control unit 71 executes in parallel with search result evaluation processing and the like will be described. FIG. 19 is a flowchart showing output processing executed by the control unit 71.

  When the output process is executed, the control unit 71 determines whether there is content that has not been output among the content stored in the storage unit 73 (S1010). The determination as to whether or not content that has not been output exists in the storage unit 73 is made based on the acquisition presence information and the output presence information in the search result management table. Specifically, when individual result information whose acquisition presence / absence information is “YES” and output presence / absence information is “NO” is registered in the search result management table, the content corresponding to this individual result information is displayed. It is determined that the content has not been output, and it is determined that the content that has not been output exists in the storage unit 73.

  As a result of the determination, if it is determined that there is no unoutput content in the storage unit 73 (No in S1010), the control unit 71 waits until unoutput content occurs, and there is unoutput content. If it is determined (Yes in S1010), one content that has the highest evaluation value of the corresponding individual result information among the unoutput content is selected as a processing target (S1015). However, if the maximum value is not greater than 0 (No in S1020), the output presence / absence information of the individual result information corresponding to the selected content is updated to “YES”, and the process proceeds to S1010. On the other hand, if the maximum value is greater than 0 (Yes in S1020), it is determined whether the content to be processed is a term search result based on the individual result information corresponding to the content (S1025).

  Specifically, for example, it is determined whether or not the content to be processed is a term search result by determining whether or not the category information indicated by the individual result information corresponding to the content is “term search”. To do. In addition, the content may be analyzed to determine whether the content to be processed is a term search result.

  If it is determined in S1025 that the content is the term search result (Yes in S1025), the control unit 71 displays the content in the term display area on the content providing screen of the display unit 77 (S1030), and then proceeds to S1060. . On the other hand, if it is determined in S1025 that the content is not a term search result, the process proceeds to S1035, where it is determined whether or not the content to be processed is a video file, and it is determined that the content to be processed is a video file. (Yes in S1035), the content to be processed and the source generation time information indicated by the source information from which the content is acquired are associated and registered in the video reproduction data queue. Thereafter, the process proceeds to S1060.

  In the information collection and analysis apparatus 70 of the present embodiment, the control unit 71 has a function as a moving image reproduction data queue and a function of reproducing content (moving image file) registered in the moving image reproduction data queue. FIG. 21A is an explanatory diagram showing a moving image reproduction data queue realized by the control unit 71 and a moving image reproduction region display control process executed by the control unit 71.

  The video playback area display control process plays back the content registered in the video playback data queue in the order of registration, outputs the video to the video playback area on the content providing screen, and outputs the sound to the sound output unit 79. .

  Returning to FIG. 19, if it is determined in S1035 that the content to be processed is not a moving image file (No in S1035), the process proceeds to S1045 to determine whether the content to be processed is an audio file. If it is determined that the file is an audio file (Yes in S1045), the content (audio file) is reproduced, and the sound output unit 79 outputs the content as sound (S1050). Thereafter, the process proceeds to S1060.

  In addition, when the reproduction process of the audio file that has started reproduction in the past is being continued at the time of executing S1050, the reproduction process of the audio file is interrupted and the above-described content (audio file) to be processed is reproduced. To do. In addition, when a moving image with sound is reproduced by the moving image reproduction area display control process, the sound output unit 79 executes sound output of the content (audio file) instead of the sound output.

  If it is determined that the content to be processed is not an audio file (No in S1045), the process proceeds to S1055, and the control unit 71 executes display control processing. FIG. 20 is a flowchart illustrating display control processing executed by the control unit 71.

  When the display control process is executed, the control unit 71 currently occupies the total content evaluated in the display data queue of the text / image area (the content selected in S1015 in FIG. 19). ) Is evaluated, and the ratio is used as a weight (S1110).

  When the processing in S1110 is completed, the display character size of the content currently being processed is set to 20 pt * weight (S1115), and the vertical size of the display area of the content in the text / image area is set to 200 mm * weight (S1120). ), The density of the display character of the content is determined by calculating the standard character * (0.5 + weight) (S1125).

  In S1130, it is determined whether or not the weight is greater than 0.5. If the weight is determined to be greater than 0.5 (Yes in S1130), the display indent of the content is set to 0 mm from the left end of the display area. (S1135) The display characters are also bold (S1140). Then, the process proceeds to S1165.

  On the other hand, if it is determined that the weight is not greater than 0.5 (No in S1130), it is determined whether the weight is less than 0.2 (S1145), and if it is determined that the weight is less than 0.2 (Yes in S1145). ), The display indent of the content is set to 20 mm from the left end of the display area (S1155), and the image information is deleted from the display target (S1160). Then, the process proceeds to S1165.

  If it is determined that the weight is not smaller than 0.2 (No in S1145), the display indent of the content is set to 10 mm from the left end of the display area (S1150), and the process proceeds to S1165.

  In step S1165, the control unit 71 registers the content and the source generation time information indicated by the source information from which the content is acquired in association with each other in the display data queue. Thereafter, the display control process ends.

  In the information collection and analysis apparatus 70 of the present embodiment, the control unit 71 has a function as a display data queue, a function of displaying content (character / image information such as a web page) registered in the display data queue, Have FIG. 21B is an explanatory diagram showing a display data queue realized by the control unit 71 and a text / image area display control process executed by the control unit 71. The text / image area display control process displays the contents registered in the display data queue in the order of registration, and displays them in the text / image area on the content providing screen.

  Returning to FIG. 19, when the display control process in S <b> 1055 is finished, the control unit 71 proceeds to S <b> 1060 and updates the output presence / absence information of the individual result information corresponding to the processing target content to “YES”, and then Then, the process proceeds to S1065, and the content in which the difference between the source generation time and the current time exceeds a predetermined time is deleted from the moving image reproduction data queue and the display data queue.

  Thereafter, it is determined whether or not an end command is input (S1070). When it is determined that no end command is input (No in S1070), the process proceeds to S1010 and when it is determined that an end command is input ( In step S1070, the output process ends.

  However, in this output process, since the content having the maximum evaluation value is preferentially selected as a processing target in S1015, the effect specific to the present invention is exhibited. Therefore, the generation speed of the unoutput content (reading process execution cycle) On the other hand, if the execution cycle of S1015 is too early, the operation of selecting the content having the maximum evaluation value as the processing target will not make sense. Therefore, in the output processing of the present embodiment, when it is determined No in S1070, until the transition timing of a predetermined cycle set sufficiently slower than the generation speed of non-output content (reading processing execution cycle) arrives. , The process waits without shifting to S1010, and after the shift timing has arrived, the process shifts to S1010 to determine whether or not unoutput content exists.

(5) Effects of the present embodiment According to the content providing system 1 of the present embodiment configured as described above, the information collection and analysis device 70 is displayed on the display unit 43 in the personal computer 30 by the source information registration process. Display target data (including data input from the user to the input unit 41 and displayed on the display unit 43) is acquired from the personal computer 30, and the digital telephone device 50 receives the data from the user to the sound input unit 61. The input data and the data output to the user from the sound output unit 63 are acquired from the digital telephone device 50 (data acquisition means).

  When the source information is registered, in S415, keywords are extracted from the received file associated with the source information for each source information (extraction means), and the keyword is an interest estimation table for each keyword. In accordance with this interest estimation table, the keyword class (kind of interest) is discriminated (S550), and the discriminated class element (class value) is incremented by 1 to obtain the user's interest. Quantify things and obtain interest vectors (first interest estimation means).

  In addition, each time the current interest table update process is executed, the information collection / analysis device 70 adds the latest interest vector at that time to the past interest vector to generate a cumulative interest vector. Based on the cumulative interest vector, the transition of the interest vector in the past predetermined period (twice) is evaluated, a transition coefficient K is calculated, and this is applied to the latest interest vector to correct the latest interest vector. As a result, the current user's main interest is obtained as the current interest vector (second interest estimation means).

  In the present embodiment, by applying the transition coefficient K to the latest interest vector, the value of the element corresponding to the type of interest that has recently been increasing is amplified, and the element corresponding to the type of interest that is calming down. And the current interest vector of the user is represented in the current interest vector.

  In the present embodiment, based on the individual search information with the latest source generation time, the content corresponding to the search key indicated in the individual search information is searched, and the search results (in the order in which the search results are obtained) Based on the individual result information), in S830, the content is acquired (content acquisition means) and stored in the storage unit 73. At this time, in the search result management table, the storage destination of the content is recorded in the content storage destination address information, and the content and the interest vector of the individual search information are associated and stored in the storage unit 73 (content recording means).

  In addition, the information collection and analysis apparatus 70 repeatedly performs the processing of S860 every time content is acquired, thereby calculating the inner product of the corresponding interest vector and the current interest vector for each content stored in the storage unit 73. Thus, the similarity between the user's interests corresponding to each content and the user's main interests is evaluated (similarity evaluation means).

  In addition, each content stored in the storage unit 73 is preferentially selected as an output target from content evaluated as having a high degree of similarity (content with a high evaluation value) (S1015), and this is selected as a content providing screen or the like. To the user (content output control means).

  As described above, the content providing system 1 according to the present embodiment evaluates the current user's interest in the content after acquiring the content, and if the interest is high, the content is transmitted to the user regardless of the source generation time. Therefore, content suitable for the user's interest is appropriately provided to the user in accordance with the user's interest rather than sequentially processing the output candidate contents in the conventional manner. Can do.

  In particular, in the present embodiment, the main interest of the current user is estimated, the degree of interest in the content is evaluated, and the output priority is determined. Therefore, it is possible to appropriately output content that is highly useful to the user. it can.

  In the present embodiment, keywords are extracted from text data obtained from the personal computer 30 and the digital telephone device 50, the user's interest is estimated, and the interest vector is calculated. Therefore, the context of the sentence indicated in the data It is possible to estimate the interest more easily than the case where the user's interest is estimated by interpreting. In addition, since interests are expressed in vectors, the similarity between multiple interests (interest vectors) can be evaluated appropriately by taking the inner product, and content that matches the user's interests can be appropriately displayed to the user. Can be provided.

  Further, according to the present embodiment, when content is output, since the evaluation result is not 0 or less (No in S1020), the output content is relatively consistent with the user's interest (that is, value) Content that is low in value for the user can be prevented from being output.

  In the above embodiment, when searching for content, processing is performed from the search key with the latest source generation time, each content corresponding to the search result is acquired in the order of generation of the search result, and the acquired content is Although the priority is output from the one that is close to the current main interest, the content is obtained by searching the content that is close to the current main interest of the user. The information collection / analysis apparatus 70 may be configured so as to be preferentially performed from those close to the main concern.

[Other Embodiments]
Hereinafter, as described above, the configuration of the information collection / analysis apparatus 70 that preferentially searches and acquires content that is close to the user's current main interest will be described. The information collection / analysis apparatus 70 executes the search control process shown in FIG. 22 instead of the search control process shown in FIG. 11, and searches the search information shown in FIG. 23 instead of the search information management table shown in FIG. Using the management table, the search process shown in FIG. 24 is executed instead of the search process shown in FIG. 13, and the read control process shown in FIG. 25 is executed instead of the search result evaluation process shown in FIG. If the information collecting / analyzing apparatus 70 of the above embodiment is configured, it will be completed. Therefore, hereinafter, only the configuration different from that of the above embodiment will be described with reference to FIGS.

(1) About Search Control Processing In this embodiment, the control unit 71 executes the search control processing shown in FIG. 22 instead of the search control processing shown in FIG. FIG. 22 is a flowchart showing search control processing executed by the control unit 71.

  When the search control process shown in FIG. 22 is executed, the control unit 71 first determines whether or not the unprocessed individual search information for the process of S1270 exists in the search information management table (S1210). If it is determined that there is no individual search information (No in S1210), the process waits until unprocessed individual search information is generated. If it is determined that there is unprocessed individual search information (Yes in S1210), the process proceeds to 1220. Then, the current interest table update process shown in FIG. 17 is executed. The contents of the current interest table update process are as described above.

  When the current interest table update process ends in S1220, the control unit 71 is unprocessed individual search information for the process in S1270, and the inner product R of the interest vector and the current interest vector registered in the current interest table. Is selected as a processing target (S1230).

  More specifically, in S1230, the normal (unit) vector E of the interest vector V of each individual search information that has not been processed in the processing of S1270 is calculated, and the normality of the current interest vector Vc registered in the current interest table is calculated. The (unit) vector Ec is calculated, and the inner product R of each vector E and the vector Ec is taken, and the individual search information corresponding to the vector E having the maximum value R is selected as a processing target.

  Thereafter, the control unit 71 proceeds to S1235, and again selects the individual search information with the latest source generation time as the processing target from the individual search information selected as the processing target in S1230. Note that this processing is meaningful when there are a plurality of pieces of individual search information having the maximum value R in S1230.

  When the processing in S1235 is completed, the control unit 71 proceeds to S1240, and once again selects one individual search information having the maximum search key from the individual search information selected as the processing target in S1235. (S1240). Note that this processing is meaningful when multiple pieces of individual search information having the same source generation time are selected in S1235.

  When the processing in S1240 is completed, the control unit 71 writes the maximum inner product R calculated in S1230 as the evaluation value of the individual search information to be processed selected in S1240 in the search information management table (S1250). In the present embodiment, the search information management table stores individual search information including source generation time information, interest vectors, search key information, search key number information, and evaluation value information. Assume that it is configured as possible.

  FIG. 23 is an explanatory diagram showing the configuration of the search information management table of this embodiment. The evaluation value information in each individual search information is set to NULL in the initial stage, and is rewritten to the inner product value R by the processing of S1250.

  When the processing in S1250 is completed, the control unit 71 selects one piece of search program information that has not been issued from the search program management table (S1260). Then, the individual search information selected in S1240 is designated as a processing target, and a search process for the selected search program information is started (S1270).

  That is, in S1260, the control unit 71 searches the search program information (not yet queried) that has not been transmitted to the URL destination information search server device 20 according to the address information from the search program information included in the search program management table shown in FIG. Issuance search program information) is selected as a processing target, and the search process is started in S1270. The activated search process is managed by the scheduler and executed in parallel with other processes, and as shown in FIG. 24, a query is generated based on the query generation information indicated by the specified search program information. This is transmitted to the information search server device 20 at the URL indicated by the address information of the search program information.

  When the processing in S1270 is completed, the control unit 71 determines whether or not search query information that has not been issued a query exists in the search program management table (S1280), and determines that search program information that has not been issued a query exists. Then (Yes in S1280), the process proceeds to S1260, where one search program information to be processed is selected from the search program information that has not been issued a query, and the processes after S1270 are executed.

  On the other hand, if it is determined that the search program information that has not been issued is not present in the search program management table (No in S1280), the control unit 71 determines whether or not an end command has been input (S1290). If it is determined that is not input (No in S1290), the process proceeds to S1210. On the other hand, if it is determined that an end command has been input (Yes in S1290), the search control process ends.

  However, in this search control process, since the individual search information with the maximum evaluation value is preferentially selected as a processing target in S1230, an effect specific to the present invention is exhibited. If the execution cycle of S1230 is too early, the operation of selecting the individual search information with the maximum evaluation value as the processing target will not make sense. Therefore, in the search control processing of this embodiment, when it is determined No in S1290, the process waits without shifting to S1210 until the shift timing of a predetermined period set sufficiently slower than the generation speed of the individual search information comes. Then, after the transition timing has arrived, the process proceeds to S1210.

(2) Search Process Next, a search process realized by each search process executed in parallel by the control unit 71 of the present embodiment will be described. FIG. 24 is a flowchart showing search processing.

  When the search process shown in FIG. 24 is executed, the control unit 71 first reads the selected search program information from the search program management table (S1305), and the query generation information included in the read search program information is designated. Based on a group of search keys included in the individual search information to be processed, a search expression in which the search key included in the individual search information is set as the search key is generated, and the information search server device 20 storing the search expression is generated. A query that is an inquiry signal is generated (S1310).

  Then, the generated query is transmitted to the URL destination indicated by the search program information, and the information search server device 20 is caused to execute the URL destination search program included in the information search server device 20 (S1315). At this time, the information search server device 20 searches the content corresponding to the search key included in the individual search information according to the search formula included in the received query, and stores the link information (URL information) of the content as the search result. The response signal is returned to the information collection and analysis device 70.

  After transmitting the query in S1315, the control unit 71 determines whether or not a response signal has been received from the information search server device 20 (S1320), and determines that the response signal has not been received (No in S1320). Wait until a response signal is received.

  On the other hand, if it is determined in S1320 that a response signal has been received (Yes in S1320), the process proceeds to S1325 to determine whether the response signal is a search result or a response signal indicating that the search has been interrupted. If it is determined that the response signal is not a search result (a response signal indicating that the search is interrupted) (No in S1325), the search process (search process) is terminated.

  On the other hand, if it is determined that the response signal is a search result (Yes in S1325), it is determined whether the category information of the designated search program information is “term search” (S1330), and “term search”. If it is determined that the search is a “general search” (No in S1330), a predetermined number of links (for example, five) for a general search are obtained from a search result including a group of link information received as a response signal. Information (URL information) is extracted, individual result information describing the corresponding link information is generated for each link information (S1335), and the storage unit 73 stores the individual result information for each generated link information. Register in the management table (S1340).

  When the processing in S1340 is completed, the process proceeds to S1360, and the evaluation value indicated by the specified individual search information to be processed is displayed in the evaluation value column of each individual result information registered in the search result management table this time. Write and update the search result management table. Thereafter, the search process (search process) is terminated.

  On the other hand, when it is determined in S1330 that the category information of the designated search program information is “term search” (Yes in S1330), the control unit 71 obtains a search result including a group of link information received as a response signal. From this, link information (URL information) for a predetermined number (for example, one) for term search is extracted, individual result information describing the link information is generated (S1345), and the generated individual result information Is registered in the search result management table stored in the storage unit 73 (S1350).

  When the process in S1350 is completed, the process proceeds to S1360, and the evaluation value indicated by the specified individual search information to be processed is displayed in the evaluation value column of the individual result information registered in the search result management table this time. Write and update the search result management table. Thereafter, the search process (search process) is terminated.

(3) Read Control Process Next, the read control process executed by the control unit 71 instead of the search result evaluation process shown in FIG. 15 will be described. FIG. 25 is a flowchart showing the read control process executed by the control unit 71.

  When the read control process shown in FIG. 25 is executed, the control unit 71 first determines whether or not unprocessed individual result information exists in the search result management table for the process of S1430 (S1410). If it is determined that the individual result information of the process does not exist (No in S1410), the process waits until the unprocessed individual result information is registered in the search result management table, and if it is determined that the unprocessed individual result information exists ( Yes in S1410), the process proceeds to S1420.

  After shifting to S1420, the control unit 71 selects one piece of unprocessed individual result information as a processing target, and then executes the reading process shown in FIG. 16 (S1430). The contents of the reading process are as described above.

  When this reading process is completed, the control unit 71 proceeds to S1440 and determines whether or not the content can be normally read in the reading process in S1430. If it is determined that the content can be read normally (Yes in S1440), the process proceeds to S1445. On the other hand, if it is determined that the content could not be read normally (No in S1440), it is determined whether an end command is input (S1450), and if it is determined that no end command is input (No in S1450). ), The process proceeds to S1410. On the other hand, when it is determined that an end command has been input (Yes in S1450), the reading control process is ended.

  In S1445, the control unit 71 updates the acquisition result information of the individual result information corresponding to the content that has been successfully read in the reading process in S1430 to “YES” and enters the acquisition completion time information area. Write the current time. Thereafter, the process proceeds to S1450, where it is determined whether or not an end command is input (S1450). When it is determined that an end command is input (Yes in S1450), the read control process is ended.

(4) Effects of the present embodiment In the present embodiment, the interest vector is calculated by the keyword extraction process, and the acquired data (text data) from the personal computer 30 or the digital telephone device 50 used to calculate the interest vector. Then, the individual search information that associates the search key extracted from the acquired data with the interest vector is registered in the search information management table (data recording means).

  In addition, the information collection and analysis apparatus 70 repeatedly executes the processing of S1230 as long as there is content-unsearched individual search information, so that the interest vector, current interest vector, Is used to evaluate the degree of similarity between the user's interests corresponding to each individual search information and the user's main interests (similarity evaluation means).

  Then, the individual search information is preferentially selected from the individual search information (individual search information having a high evaluation value) evaluated as having a high similarity (S1230 to S1240), and the search indicated by the individual search information The content corresponding to the key is searched (S1270), and the corresponding content is acquired from the web server based on the search result (individual result information) in the reading control process in the order in which the search is completed (content acquisition control means). .

  That is, in the content providing system 1 of the present embodiment, content having a high similarity to the user's main interest is preferentially acquired from the web server and output to the user regardless of the source generation time. According to the present embodiment, in order to preferentially acquire highly useful content as described above, the user can appropriately and appropriately reduce the load on the CPU and the load on the network when acquiring the content. Content that matches the interest of the user can be acquired and provided to the user.

  In addition, in this embodiment, as described above, the transition coefficient K is applied to the latest interest vector to amplify the value of the element corresponding to the type of interest that has recently been increasing, and the type of interest that is still calming down. The value of the element corresponding to is reduced, and the main interest of the user at present is quantified and used as the reference for the evaluation.

  Therefore, according to the present embodiment, it is possible to appropriately search / acquire content that matches the current user's main interests, and to appropriately provide highly useful content to the user.

[Others]
In the content providing system 1 of the above embodiment, the information collection and analysis device 70 exists as a single device, but such a configuration is merely a configuration example for facilitating the description, and other than this. It is possible to adopt various configurations. For example, a configuration in which the personal computer 30 and the information collection / analysis apparatus 70 are integrated may be employed.

  In the above embodiment, as shown in FIG. 8, keywords are extracted from the acquired data from the personal computer 30 or the digital telephone device 50, and the number of keywords that match the keywords registered in the interest estimation table Although the interest is quantified and expressed as an interest vector, the user's interest may be quantified by other methods.

  For example, in the interest estimation table, the type of interest and the score are determined for each keyword. If the same keyword is included in the acquired data from the personal computer 30 or the digital telephone device 50, the interest The user's interests may be quantified by counting the number of types.

It is explanatory drawing showing the structure of the content provision system 1 to which this invention was applied. 4 is a flowchart showing a personal computer monitoring process executed by a monitoring unit 35 of the personal computer 30. 4 is a flowchart showing telephone monitoring processing repeatedly executed by a monitoring unit 55 of the digital telephone device 50. It is explanatory drawing showing the structure of the content provision screen which the display part 77 displays. It is a flowchart showing the source information registration process which the control part 71 performs. It is explanatory drawing showing the structure of a source management table. It is a flowchart showing the keyword extraction process which the control part 71 performs. It is a flowchart showing the interest vector calculation process which the control part 71 performs. It is explanatory drawing showing the structure of an interest estimation table. It is explanatory drawing showing the structure of a search information management table. It is a flowchart showing the search control process which the control part 71 performs. It is explanatory drawing showing the structure of a search program management table. It is a flowchart showing the search process implement | achieved in a search process. It is explanatory drawing showing the structure of a search result management table. It is a flowchart showing the search result evaluation process which the control part 71 performs. It is a flowchart showing the read process implement | achieved in the read process. It is a flowchart showing the current interest table update process which the control part 71 performs. It is explanatory drawing (a) showing the structure of an interest history table, and explanatory drawing (b) showing the structure of a current interest table. It is a flowchart showing the output process which the control part 71 performs. It is a flowchart showing the display control process which the control part 71 performs. It is explanatory drawing (a) which showed the moving image reproduction data queue and the moving image reproduction area display control process, and explanatory drawing (b) which showed the display data queue and the text / image area display control process. It is a flowchart showing the search control process in other embodiment. It is a flowchart showing the structure of the search information management table in other embodiment. It is a flowchart showing the search process in other embodiment. It is a flowchart showing the read-out control process which the control part 71 performs by other embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Content provision system, 10 ... Web server, 20 ... Information retrieval server apparatus, 21, 31, 51, 71 ... Control part, 23, 33, 53, 73 ... Storage part, 25, 37, 57, 75 ... Communication part , 30 ... PC, 35, 55 ... monitoring unit, 40, 60 ... user interface, 41 ... input unit, 43, 77 ... display unit, 50 ... digital telephone device, 61 ... sound input unit, 63, 79 ... sound output unit , 70 ... Information collection and analysis device

Claims (8)

In an information processing apparatus comprising a user interface, data acquisition means for acquiring from the information processing apparatus at least one of data input to the user interface from a user and data output to the user by the user interface;
Each time the data acquisition means acquires the data, first interest estimation means for estimating the user's interest based on the acquired data;
A first estimation of the main interest of the user at present is performed by correcting the latest estimation result by the first interest estimation means based on the transition of the estimation result for the past predetermined period by the first interest estimation means. Dual interest estimation means;
Content acquisition means for acquiring content corresponding to each data acquired from the information processing apparatus by the data acquisition means from an apparatus holding the content;
Content recording in which the content acquired by the content acquisition unit and the estimation result of the first interest estimation unit obtained from the acquired data from the information processing apparatus corresponding to the content are associated and stored in the storage unit Means,
For each content stored in the storage unit, an estimation result of the first interest estimation unit stored in the storage unit in association with the content, and a main interest of the user estimated by the second interest estimation unit A similarity evaluation means for evaluating the similarity,
Output means for outputting each content stored in the storage means to the user;
Content output control means for preferentially outputting each content stored in the storage means from the content evaluated by the similarity evaluation means as having a high similarity, to the output means;
A content output system comprising: Extraction means for extracting a keyword from the obtained data every time the data obtaining means obtains the data;
With
The first interest estimation unit is configured to estimate the interest of the user based on the keyword extracted by the extraction unit from the acquired data every time the data acquisition unit acquires the data.
The content acquisition unit is configured to acquire, for each data acquired by the data acquisition unit from the information processing device, content related to the keyword extracted by the extraction unit from the data from a device holding the content. The content output system according to claim 1, wherein: In an information processing apparatus comprising a user interface, data acquisition means for acquiring from the information processing apparatus at least one of data input to the user interface from a user and data output to the user by the user interface;
Each time the data acquisition means acquires the data, first interest estimation means for estimating the user's interest based on the acquired data;
A data recording means for associating the estimation result of the first interest estimation means with the acquired data from the information processing device used to derive the estimation result, and storing the data in the storage means;
A first estimation of the main interest of the user at present is performed by correcting the latest estimation result by the first interest estimation means based on the transition of the estimation result for the past predetermined period by the first interest estimation means. Dual interest estimation means;
Content acquisition means for acquiring content corresponding to each acquisition data from the information processing apparatus stored in the storage means from an apparatus holding the content;
Output means for outputting each content acquired by the content acquisition means to the user;
For each acquired data stored in the storage means, the estimation result of the first interest estimation means stored in the storage means in association with it, and the main interest of the user estimated by the second interest estimation means A similarity evaluation means for evaluating the similarity with
Each acquisition data stored in the storage means is preferentially selected as a processing target from data evaluated as having a high similarity by the similarity evaluation means, and corresponds to the selected acquisition data of the processing target Content acquisition control means for causing the content acquisition means to acquire content preferentially;
A content output system comprising: Extraction means for extracting a keyword from the obtained data every time the data obtaining means obtains the data;
With
The first interest estimation means estimates the user's interest each time the data acquisition means acquires the data, based on the keyword extracted by the extraction means from the acquired data,
The data recording means is an estimation result of the first interest estimation means, acquired data from the information processing device used to derive the estimation result, a keyword extracted by the extraction means, Are stored in the storage means,
The said content acquisition means is comprised so that the content relevant to the keyword extracted from each said acquisition data which the said memory | storage means memorize | stores may be acquired from the apparatus holding content. Content output system.   Each time the data acquisition unit acquires the data, the first interest estimation unit classifies each keyword extracted by the extraction unit from the acquired data into one of a plurality of predetermined classes. The class value for each class is obtained based on a keyword, and based on the class value, the user's interests are expressed as vectors, and the user's interests are estimated. Or the content output system of Claim 4.   The second interest estimation means obtains a change amount of the estimation result expressed in the vector for the past predetermined period by the first interest estimation means, and uses the change amount, the first interest estimation means 6. The content output system according to claim 5, wherein the content output system is configured to estimate the current main interest of the user by correcting the latest estimation result. In a computer having storage means,
In an information processing apparatus comprising a user interface, data acquisition means for acquiring from the information processing apparatus at least one of data input to the user interface from a user and data output to the user by the user interface;
Each time the data acquisition means acquires the data, first interest estimation means for estimating the user's interest based on the acquired data;
A first estimation of the main interest of the user at present is performed by correcting the latest estimation result by the first interest estimation means based on the transition of the estimation result for the past predetermined period by the first interest estimation means. Dual interest estimation means;
Content acquisition means for acquiring content corresponding to each data acquired from the information processing apparatus by the data acquisition means from an apparatus holding the content;
Content to be stored in the storage unit in association with the content acquired by the content acquisition unit and the estimation result of the first interest estimation unit obtained from the acquired data from the information processing apparatus corresponding to the content Recording means;
For each content stored in the storage unit, an estimation result of the first interest estimation unit stored in the storage unit in association with the content, and a main interest of the user estimated by the second interest estimation unit A similarity evaluation means for evaluating the similarity,
Output means for outputting each content stored in the storage means to the user;
A program for realizing a function as a content output control unit that preferentially outputs each content stored in the storage unit from the content evaluated as having a high similarity by the similarity evaluation unit to the output unit . In a computer having storage means,
In an information processing apparatus comprising a user interface, data acquisition means for acquiring from the information processing apparatus at least one of data input to the user interface from a user and data output to the user by the user interface;
Each time the data acquisition means acquires the data, first interest estimation means for estimating the user's interest based on the acquired data;
A data recording means for associating and storing the estimation result of the first interest estimation means and the acquired data from the information processing apparatus used to derive the estimation result in the storage means;
A first estimation of the main interest of the user at present is performed by correcting the latest estimation result by the first interest estimation means based on the transition of the estimation result for the past predetermined period by the first interest estimation means. Dual interest estimation means;
Content acquisition means for acquiring content corresponding to each acquisition data from the information processing apparatus stored in the storage means from an apparatus holding the content;
Output means for outputting each content acquired by the content acquisition means to the user;
For each acquired data stored in the storage means, the estimation result of the first interest estimation means stored in the storage means in association with it, and the main interest of the user estimated by the second interest estimation means A similarity evaluation means for evaluating the similarity with
Each acquisition data stored in the storage means is preferentially selected as a processing target from data evaluated as having a high similarity by the similarity evaluation means, and corresponds to the selected acquisition data of the processing target A program for realizing a function as content acquisition control means for causing the content acquisition means to acquire content preferentially.

Images