JP4396444B2 - Phrase extraction device and program - Google Patents

Description

The present invention relates to a phrase extractor及 beauty programs, in particular, in providing topic of events to start at a time, such as television programs and radio programs, a network such as the Internet document information described for the program obtained from one or more information providing servers connected to, extracts phrases become the subject in response to the attention of the program, regarding the phrase extractor及 beauty program for providing.

  In recent years, with the development of computer networks such as the Internet, a large amount of electronic document information has been accumulated one after another. In particular, it has become possible for individuals to easily send their opinions and opinions on matters of interest using bulletin boards and blog services. Therefore, by collecting and analyzing a lot of information that is sent one after another, such as news, bulletin boards, and blogs, it is possible to grasp the latest news and events.

  Conventionally, there has been proposed an information flow search method in which information transmitted from a plurality of information providing servers is classified into categories for each topic, and the time transition of the topic is presented and searched. In this information flow search method, processing is performed in which similar documents are collected from a document set transmitted in a certain period and assigned to a category (for example, see Patent Document 1). For this reason, since a category is assigned only after a plurality of similar documents are transmitted, the topic can be presented only after a lapse of time since the information on the topic is transmitted for the first time.

  In addition, the personal interest includes an event that starts at a specific date and time such as a program. Usually, after experiencing the event, the impression and opinion are described and transmitted. Therefore, the topic can be presented only after the program ends.

Currently, there is a case where a user writes on a bulletin board or the like while watching a program, but a topic cannot be presented when there is only a small amount of document information.
JP 2000-242652 A

  As described above, in the conventional technique, a topic can be provided only after the program has already ended, and the latest program information that may become a topic cannot be provided quickly.

The present invention has been made in view of the above points, and it is possible to extract a phrase that allows a user to provide an interesting phrase for an event that starts at a specific time before the event is started. and to provide a device及 beauty program.

  FIG. 1 is a diagram for explaining the principle of the present invention.

This onset Ming, a topic related to events that are start at a specific time, such as television programs and radio programs, and document information that has been published in the network, the schedule of the program is to provide extracts from the program table that has been recorded In the phrase extraction method in the phrase extraction device used for
A program that searches and extracts a document including a program title from a set of documents having time information read from a document information storage unit that stores documents having time information, and stores the program in a document information storage unit for each program. A separate document extraction step (step 1);
From the extracted documents , documents having time information within a predetermined attention level calculation period corresponding to the broadcast period of the program from the program start date and time are totaled for each program, and the program attention level is set as the program attention level. Attention level calculation step (step 2) for each program stored in the table,
When the program attention level table is referenced, a program with a high degree of attention is selected in the previous broadcast, and the next broadcast of the program is within a preset time before the broadcast of the program until the broadcast start time Acquires the subtitle or outline of the program from the program guide storage means storing the program guide, performs morphological analysis, extracts words and phrases based on the appearance position and number of characters of the phrases, and starts broadcasting after the program broadcasts If it is within the preset time from the time, the document stored in the program-specific document information storage means is subjected to morphological analysis, the number of documents appearing for each phrase is obtained, and the phrases having a large number of appearing documents are extracted. And a word extraction step (step 3).

This onset Ming, in the program-specific attention degree calculation step (step 2), the search request search term to the attention degree calculation period, using the information specifying the user who has entered the search term, the same user is short by counting and once when you enter multiple times the same keyword in the time interval, aggregate search terms using the number of which related to the turn of sets, the sum of the aggregate value of the document number as the target of the program Store in the program attention level table .

This onset Ming, in the program-specific attention degree calculating step (step 2), to set a high degree of attention for the new program.

This onset Ming, in the phrase extraction step (Step 3),
The maximum period until the start of broadcasting is set in advance according to the program broadcasting cycle,
The start date and time of the program to be broadcast and the period from the current date and time to the start of broadcast are obtained, and words and phrases are extracted for the program whose period until the start of the broadcast does not exceed the maximum period.

  FIG. 2 is a principle configuration diagram of the present invention.

The present invention (Claim 1 ) extracts a topic related to an event that starts at a specific time such as a television program or a radio program from document information published on the network or a program table on which a program schedule is recorded. A phrase extraction device used to provide
A document including a program title is searched and extracted from a set of documents having time information read from the document information storage unit 202 that stores documents having time information, and stored in the program-specific document information storage unit 103 for each program. A program-specific document extracting means 101,
From the extracted documents , documents having time information within a predetermined attention level calculation period corresponding to the broadcast period of the program from the program start date and time are totaled for each program, and the program attention level is set as the program attention level. Attention level calculation means 102 for each program stored in the table;
When the program attention level table is referenced, a program with a high degree of attention is selected in the previous broadcast, and the next broadcast of the program is within a preset time before the broadcast of the program until the broadcast start time The subtitle or outline of the program is acquired from the program guide storage means 203 in which the program guide is stored, morphological analysis is performed, the phrase is extracted based on the appearance position and the number of characters of the phrase, and the broadcast starts after the program is broadcast If it is within a preset time from the time, the morphological analysis is performed on the document stored in the program-specific document information storage unit 103 to obtain the number of documents that appear for each word, and the words with a large number of appearing documents are extracted. And phrase extracting means 104.

According to the present invention (claim 2 ), the program-specific attention level calculating means 102
When the same user inputs the same keyword a plurality of times at short time intervals using the information for specifying the user who has input the search word from the search words requested during the attention degree calculation period , it is counted once. by aggregates search terms occupancy of associated turn sets includes means for storing the program of interest degree table the sum of the aggregate number of documents as a target of the program.

In the present invention (claim 3 ), the attention level calculation means 102 for each program sets a high attention level for a new program.

According to the present invention (claim 4 ), in the phrase extracting means 104,
The maximum period until the start of broadcasting is set in advance according to the program broadcasting cycle,
Means for obtaining a start date and time of a program to be broadcast and a period from the current date and time to the start of the broadcast, and extracting a phrase for a program whose period until the start of the broadcast does not exceed the maximum period.

The present invention (Claim 5 ) is a phrase extraction program for causing a computer to function as each means constituting the phrase extraction apparatus according to any one of Claims 1 to 4 .

  As described above, according to the present invention, the degree of attention of a program that starts at a specific time such as a TV program or a radio program is calculated, and the program that is to be started is calculated from the program table that introduces the contents. By extracting the phrase, it is possible to provide information that may become a topic before the program is started.

  In addition, after the program is started, it is possible to extract and provide a topical phrase from document information transmitted by people watching the program.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 3 is a block diagram of the phrase extracting device in one embodiment of the present invention.

  The word extraction device 100 shown in the figure is connected to a search word storage unit 201, a document information storage unit 202, a program information storage unit 203, and a word storage unit 204 as external storage devices. The information from the storage unit 202 and the program information storage unit 203 is input, and the extracted phrase is output to the phrase storage unit 204.

  The phrase extraction device 100 includes a program information extraction unit 101, a program attention calculation / storage unit 102, a program-specific document information storage unit 103, and a phrase extraction unit 104. The program information extraction unit 101 includes a search word storage unit 201. The document information storage unit 202 is connected, and the program attention level calculation / storage unit 102 is connected to the program information storage unit 203.

  The search word storage unit 201 collects search words input by a user at a search site that provides a search service on the Internet and stores them together with the time when the search is requested. Search terms are words that are input as a clue to search for information that the user is interested in, so by collecting and analyzing the search terms regularly, you can know the information that the user is highly interested in Can do.

  The document information storage unit 202 collects newly generated document information or updated document information from document information published on the Internet such as a home page, blog, bulletin board, etc. Accumulated with information. From these document information, it is possible to collect impressions and the like of programs viewed by people who have watched the programs. In addition, when a TV station provides a program homepage, the latest contents of the program can be collected.

  The program information storage unit 203 stores information on the TV program guide. The TV program guide is composed of information related to programs such as broadcast day, day of the week, genre, program start time, program end time, program title, subtitle, and summary text, and the same program title is stored separately for each broadcast date. ing. It is assumed that the program information storage unit 203 stores information on the TV program guide for about one week ahead.

  The program attention degree calculation / accumulation unit 102 of the phrase extracting device 100 periodically acquires program information from the program information accumulation unit 203 and manages it by program title. FIG. 4 shows an example of the program totalization result table 500 and the program attention level table 510 stored in the program attention level calculation / accumulation unit 102. FIG. 4A is a configuration example of the program-by-program total result table 500. The latest program information is acquired from the program information storage unit 203, and the program information is managed by program title. The program-by-program total result table 500 includes a program ID 501, a program title 502, a cycle 503, a start date and time 504, an end date and time 505, an attention degree calculation date and time 506, and total data 507.

  The program ID 501 is a unique number for identifying the program, and the program title 502 is the program name of the television program. Programs with the same program title 502 are managed with the same program ID. In the period 503, when the TV program is a regular program (every week, every day), the number of days is recorded (for example, a program broadcast every week is a seven day period, and “7” is recorded). ing). The start date and time 504 is the broadcast start date and time of the program, the end date and time 505 is the broadcast end date and time of the program, and the attention level calculation date and time 506 is a value obtained by adding a period corresponding to the cycle 503 to the start date and time 504. Indicates the date and time to be calculated. The total data 507 is data used when calculating the degree of attention of a program, and data for a certain period can be accumulated for a specific period. Also, a ranking of program attention as shown in FIG. 4B is generated from the value of the total data 507 and accumulated in the program attention degree table 510. The program attention level table 510 stores rank 511, program ID 512, and attention level 513.

  When a program is broadcast regularly, it is updated to the next broadcast time after the attention level is calculated. For example, in FIG. 4A, if a program with the program title “Information A” with a program ID “0000101” is a program that starts broadcasting at 18:00 and ends at 18:27 every day, the degree of attention is calculated. When the date and time “January 21, 2004, 12:00: 00” has passed, it is updated to the next broadcast time (start time, January 21, 2004, 18:00, end time, January 21, 2004, 18:27). The

  The program information extraction unit 101 determines a specific period depending on how often the information related to the television program is to be provided to the user, independently of the processing in the program attention level calculation / accumulation unit 102, and the predetermined specific period Process every other.

  First, the program information extraction unit 101 acquires a program title from the program attention level calculation / accumulation unit 102, and totals the search terms that match the program title for each program title from the search terms acquired from the search term storage unit 201. Further, the document information including the program title is totaled for each program title from the document information acquired from the document information storage unit 202. Then, the program title total value calculated from the search term and the program title total value calculated from the document information are added and recorded in the program total calculation result table 500 of the program attention degree calculation / accumulation unit 102. Also, document information related to the program is organized for each program and recorded in the program-specific document information storage unit 103.

  After the processing of the program information extraction unit 101 ends, the word / phrase extraction unit 104 acquires a program title with a high degree of attention from the program attention level calculation / accumulation unit 102, and the start time of the program with a high degree of attention falls within a preset time. When the program is started, the program subtitle and the summary sentence are acquired from the program information storage unit 203, and a topical phrase is extracted and output to the phrase storage unit 204. In addition, after the broadcast of a program with a high degree of attention is started, document information is acquired from the program-specific document information storage unit 103, a phrase representing a topic is extracted, and is output to the phrase storage unit 204.

  FIG. 5 shows an example of storage of words output to the word storage unit 204. The phrase 903 extracted by the phrase extraction unit 104 is stored in the phrase storage unit 204 in association with the program title 901 and the program start date 902.

  As described above, by periodically executing the processing of the program information extraction unit 101 and the phrase extraction unit 104, a program with a high degree of attention and a phrase representing the topic of the program are sequentially output to the phrase storage unit 204. The Since the program start date and time 902 is recorded in the word / phrase accumulating unit 204, what kind of word / phrase is being talked about in the current program being broadcast, or what program will be broadcast in the future It is possible to provide information such as.

  Next, the processing flow of the program information extraction unit 101 will be described with reference to FIG.

  FIG. 6 is a flowchart of the process of the program information extraction unit in one embodiment of the present invention.

  The processing of the program information extraction unit 101 is executed periodically (for example, every 15 minutes). First, a program title is acquired from the program attention level calculation / accumulation unit 102 (step 101). If the processing is executed every 15 minutes, the time range specified here is 15 minutes before the current time, and the time range specified next time is 15 minutes continuous to the time range specified this time.

  Next, a search word is acquired by designating a time range from the search word storage unit 201 (step 102), and search words that match the program title are tabulated for each program title (step 103). It is conceivable that the same user inputs the search term multiple times. Therefore, if the same user inputs the same keyword multiple times at short time intervals based on information that identifies the user who entered the search word (for example, cookie information of the Web browser), the search is counted as one The number of input persons for each word is totaled (step 104).

  Next, the document information is acquired from the document information storage unit 202 by specifying the time range, the document information related to the program is selected, and the number of documents is totaled for each program title. The time range specified here is the same time range as the time range specified in step 102.

  Document information related to the program is organized for each program and recorded in the program-specific document information storage unit 103 (step 105). As a method of selecting a document related to a program, there is a method of searching for document information including a program title. This method is effective in that the program title can be processed at high speed when the program title is not generally used in another context (for example, “Pocket Monster”). However, since the program title is often omitted (for example, “Pocket Monster” is often abbreviated as “Pokemon”), the program title abbreviation and alias are also stored in the program attention calculation / accumulation unit 102. As a result, more document information can be selected.

  Further, when the program title is used in a context other than the program, there is a problem that document information not related to the program is mixed if the document information is selected only under the condition that the program title is included. Document information related to a program often includes words such as “view”, “broadcast”, “say”, “do”, etc. in the vicinity of the program title. It can be selected as related document information.

  As an accumulation method, an area for storing document information is set for each program ID, and the document information is accumulated by adding a file name including the time so that the correspondence with the time information can be taken in the area. .

  Then, for each program title, the total value of the search terms and the total number of documents counted from the document information are added and recorded in the total data 507 of the program total calculation result table 500 of the program attention degree calculation / accumulation unit 102 (step) 106). The total value of the search word is the number of users who requested the search by the program title, and can be said to be the number of people who are interested in the program. Also, the total number of documents represents the number of people who are interested in the program from the point that the comments and opinions about the program are documented. That is, the total data 507 stores the number of people interested in the program for each period.

  Next, the processing in the program attention level calculation / accumulation unit 102 will be described separately for management processing of the program total result table 500 and program attention level calculation / processing. First, the management process of the program-by-program total result table 500 will be described with reference to FIG.

  FIG. 7 is a flowchart of the management processing of the program-by-program total result table according to the embodiment of the present invention.

  The program attention degree calculation / accumulation unit 102 periodically acquires a program information set starting from the program information accumulation unit 203 during a designated period (step 201). If the information registration / update of the program information storage unit 203 is performed once a day, it may be executed once a day after the information registration / update process of the program information storage unit 203. Further, the program information to be started is designated such that the designated period is within 120 hours (5 days) from the current time. One program is extracted from the acquired program information set in order of oldest start time, and the following processing is performed.

  First, the program attention level calculation / accumulation unit 102 takes out a program title from the program information set and checks whether it exists in the program title 502 of the program total result table 500 (step 202). If it does not exist, it is information on a new program and is added to the program-specific storage result table 500 (step 203). At this time, since the period information of the new program is unknown, “0” is recorded. When a program title already exists in the program title 502, it is checked whether the cycle 503 of the program total table 500 is “0” (“Yes”) (step 204, Yes), and stored in the program total table 500. The program start date and time 504 and the program start date and time are extracted from the program information set, and the cycle is calculated to update the cycle 503 (step 205).

  Subsequently, it is checked whether or not the attention level has been calculated (step 206). When the attention level has already been calculated (step 206, Yes), the start date and time 504, the end date and time 505, and the attention level calculation date and time 506 are updated. (Step 207). The attention level calculation date and time 506 is recorded by adding the attention level calculation period corresponding to the cycle 503 to the start date and time 504. The attention level calculation period is a period from when a program is broadcast until it is talked about among the people and then until the topic is settled, and is obtained in advance by examining the transition of past data. For example, a value such as 18 hours for a program with a cycle of 1 day and 72 hours for a program with a cycle of 7 days is set in advance. Also, the attention level calculation period of a new program or a program that is broadcast only once is set in advance. Thus, by setting the attention level calculation period according to the start date and time and the period of the program, it is possible to calculate the attention level for each broadcast.

  The program with the program ID “0000101” with the program title “information A” has a start date “January 20, 2004, 18:00”, a period of “1” day, and a period of interest calculation period of the program with a period of one day = 18 hours, the attention degree calculation date and time is “January 21, 2004, 12:00:00”.

  For example, when a plurality of programs broadcast every day are acquired in step 201, a program having the oldest start time among the acquired programs is calculated after the attention level calculation for the previous broadcast is completed. The program recorded at step 207 and having the next oldest start time is not recorded until the interest level calculation is completed at step 206.

  Further, as a method for checking whether or not the attention level has been calculated, the attention level calculation date and time 506 is emptied when the attention level calculation process is executed. As another method, there is a method of adding an item indicating whether or not the attention level has been calculated to the program-by-program total result table 500.

  For all the program information sets acquired from the program information storage unit 203, the processing from step 202 to step 207 is repeated until the processing is completed, and when all the programs have been processed (step 208, Yes), the series of processing is completed. To do.

  Next, the flow of program attention calculation / processing will be described with reference to FIG.

  FIG. 8 is a flowchart of program attention level calculation / processing in an embodiment of the present invention.

  The program attention level calculation / storage unit 102 performs the program attention level calculation / processing for the purpose of calculating the attention level for one broadcast for each program title, and the calculated attention level is the program of the same program title next time. Used when is broadcast. Since television programs are often configured in units of 30 minutes or 1 hour, processing is executed at 30-minute intervals. There is no problem if it is executed at a time interval other than 30 minutes, but it is necessary to calculate the degree of attention by the next broadcast start date and time.

  The program attention level calculation / accumulation unit 102 searches the program total table 500 for programs whose attention level calculation date and time 506 is older than the current date and time, and acquires a set of program IDs 501 every 30 minutes (step 301). One program ID is taken out from the set of program IDs, and the total data from the start date / time 504 to the attention level calculation date / time 506 corresponding to the program ID is added to obtain the attention level. When the attention level calculation is completed, the attention level calculation date is changed to empty (step 302).

  Next, the program information storage unit 203 is searched by the program title (step 303), and when there is program information of the next broadcast (step 304, Yes), the start date and time 504 and the end date and time 505 are acquired, and the attention degree calculation date and time 506 is set (step 305).

  When all of the program ID sets acquired in step 301 have been processed (step 306, Yes), the program ID attention levels are rearranged in descending order, ranked, and stored in the program attention level table 510 (step 307). . Here, the accumulated attention level is a value indicating how many people are interested in one broadcast for each program title. The degree of attention calculated by the above processing is only a program having periodicity, and the degree of attention of a new program or a program that is broadcast only once cannot be calculated. Therefore, a new program or a program that is broadcast only once is inserted at the top of the rank calculated in step 307 (step 308).

  Thus, by increasing the degree of attention of a new program or a program that is broadcast only once, it becomes possible to extract and provide topics from document information described by people who have watched the program. Since the attention level of the program is calculated for each broadcast, the current rank does not affect the next time.

  Next, the processing flow of the phrase extraction unit 104 will be described with reference to FIG.

  FIG. 9 is a flowchart of the processing of the phrase extraction unit in one embodiment of the present invention.

  The processing by the phrase extraction unit 104 is executed after the processing of the program information extraction unit 101 is completed. That is, if the process of the program information extraction unit 101 is every 15 minutes, the process of the phrase extraction unit 103 is also executed every 15 minutes.

  The top N program IDs 512 with the highest attention level are periodically acquired from the program attention level table 510 of the program attention level calculation / accumulation unit 102, and the program title 502 corresponding to the program ID with the higher level of attention is obtained from the program total result table 500. , Start date 504, attention calculation date 506, and period 503 are acquired (step 401). One program ID, start date and time, attention level calculation date and time, and one cycle are extracted (step 402), and it is checked whether the current date and time is newer than the start date and time and before the attention level calculation date and time is satisfied (step 403). The attention degree calculation date / time is a date / time obtained by adding a period set in accordance with the cycle to the start date / time, and is a date / time when a topic breaks down among people after the program is broadcast. Therefore, when the condition of step 403 is satisfied (step 403, Yes), document information described by people is acquired from the program-specific document information storage unit 103, and a phrase representing a topic is extracted. The storage area of the program-specific document information storage unit 103 is specified by the program ID, and the document information from the start date and time to the current date is acquired in association with the time information (step 404).

  A phrase representing a topic is extracted from the acquired document information, and the program title and the start date / time are associated with each other and output to the phrase storage unit 204.

When the condition of step 403 is not satisfied (step 403, No), a period T is set to select a program to be broadcast from now on, and a program to be broadcast in the near future (step 406). The period T is set to a value smaller than (period-attention level calculation period-interval of attention level calculation process) in the case of a program broadcast periodically. If the interval of interest level calculation processing is 30 minutes as shown in the above-described example of program attention level calculation and processing, for example, in the case of a program broadcast every day, the attention level calculation period is 18 hours.
24-18-0.5 = 5.5
For programs that are broadcast less than an hour and weekly,
168-72-0.5 = 95.5
Set a value less than the time. Then, a program whose start date and time is newer than the current date and time and older than the current date and time + period T is selected (step 407, Yes), and the program information subtitle of the program information whose start date and time match from the program information storage unit 203, or Get summary statement. Since the subtitle is a word that accurately represents the program content, the subtitle is easier to convey the program content than the summary text. Therefore, if there is a subtitle, the subtitle is acquired, and if there is no subtitle, a summary sentence is acquired (step 408).

  A phrase is extracted from the acquired information, and the program title is associated with the start date and time and output to the phrase storage unit 204. The period T affects how long before the broadcast date and time information on programs with a high degree of attention can be provided. For example, when the period T is set to 3 hours, it is possible to provide information on a program with a high degree of attention that starts within 3 hours from now. In addition, when the period T is set to 24 hours for a program broadcast every week, a program with a high degree of attention to the user, “Are there any programs that are attracting attention at this time of tomorrow?” Can be provided.

  Processing is performed for all the N programs acquired in step 401, and when all the programs have been processed (step 410, Yes), a series of processing ends.

  Next, FIG. 10 illustrates an example of the start timing of each process in FIGS. 6, 8, and 9, taking the program “information A” with the program ID “0000101” as an example.

  The program ID “0000101” is a program that is broadcast every day because the start date and time 504 (a) is “January 20, 2004 18: 00: 00: 00” and the period 503 is “1” from FIG. At step 207 in FIG. 7, the attention level calculation date and time has already been calculated from the start date and time 504 and the period 503 (in the case of a program broadcast every day, if the attention level calculation period is 18 hours, the attention level calculation date and time 506 (A) is “January 21, 2004, 12:00”). If the program “information A” is a program with a high degree of attention, the processing of the program information extraction unit 101 (FIG. 6) and the processing of the word extraction unit 104 (FIG. 9 are executed every 15 minutes, for example. 4 are sequentially output every 15 minutes, and the total value is accumulated in the total data 507 in FIG.

  When the program attention level calculation / processing (FIG. 8) is executed for the first time after the attention level calculation date “January 21, 2004 12:00: 00”, the start date 504 (a) “January 20, 2004 Date of interest calculation date and time 506 (a) “Total date 507 from January 1st, 2004 at 12:00” is added and recorded in the program attention level table 510 (step 302). ).

  Further, the program information storage unit 203 is searched based on the program title, and the next program information (start date and time (b) “January 21, 2004 18:00:00”, end date and time (b) “2004 1 If it can be acquired, the start date and time and the end date and time are updated, the next attention level calculation date and time (b) is calculated, and recorded in the program-by-program total result table 500 (step 305). .

  From the previous start date / time (a) “January 20, 2004 18: 00: 00: 00” to the attention degree calculation date / time (a) “January 21, 2004 12: 00: 00: 00”, the processing of the phrase extraction unit 104 A phrase is extracted from the document information storage unit 103 in accordance with the condition of step 403 (steps 404 and 405).

  When the next start date and time “January 21, 2004, 18:00:00” period T = 3 hours ago, the condition of step 407 is met and the phrase is extracted from the program information storage unit 203 (steps 408 and 409). .

  With respect to a program that is broadcast periodically in this way, a program that has a high level of attention in the previous broadcast can be provided with a word that will become a topic before the broadcast.

  More specifically, the processing flow of the phrase extraction unit 104 will be described using the accumulation example of the program attention degree calculation / accumulation unit 102 of FIG.

  The following explanation will be given on the assumption that the current date and time is “January 20, 2004 19:15”, and the program IDs with high attention are “0000103” and “0000105”.

  First, the program ID “0000103” and the start date and time “January 20, 2004, 19:00:00” and the attention degree calculation date and time “January 23, 2004, 19:00:00” are taken out from the program total result table 500 ( Step 402). Whether the program ID 501 “0000103” satisfies the condition of Step 403 (start date and time ≦ current date and time of interest calculation date and time) is checked, and in order to satisfy the condition, document information is acquired from the program-specific document information storage unit 103 (Step 404). The phrase is extracted from the document information, and the program title “Animation C” and the start date “January 20, 2004, 19:00:00” are associated with each other and output to the phrase storage unit 204 (step 405).

Next, the program ID “0000105” and the start date “January 20, 2004 21:00:00” are extracted. It is checked whether or not the condition of step 403 is satisfied. Since the condition is not satisfied, a period T is set in step 406. Here, it is assumed that the period T is preset to 3 hours for a program broadcast every day and 24 hours for a program broadcast every week. Since the program ID 501 “0000105” has a period 503 of “7” and is broadcast every week , the period T is set to 24 hours (step 406), and the condition of step 407 (current date <start date <current date + T) Find out if it meets. In order to satisfy the condition of step 407, a summary text of the program is acquired from the program information storage unit 203 (step 408). Phrases are extracted from the summary sentence, and the program title 502 “Drama E” and the start time 504 “January 20, 2004, 21:00:00” are associated and output to the phrase storage unit 204 (step 409).

  Next, the phrase extraction method in step 405 will be described in more detail.

  In step 405, morpheme analysis processing is performed on each document content of the document information set acquired in step 404, and the document contents are decomposed into morphemes such as individual parts of speech and punctuation marks. From the decomposed morphemes, nouns, compound nouns in which a plurality of nouns are continuous, and noun phrases that function in the same way as nouns are extracted for the entire consecutive words such as “Koizumi administration support rate”. This is because nouns, compound nouns, and noun phrases are more suitable than verbs and adjectives as words representing topics. In the following description, nouns, compound nouns, and noun phrases are collectively referred to as words. After extracting words from all document information, the number of documents appearing for each word is totaled. Here, the words included in the title are excluded from aggregation. If an abbreviation or alias name of the program title is used in step 105, the abbreviation or alias name is also excluded from the total. If an abbreviation or alias name of the program title is used in step 105, the abbreviation or alias name is also excluded from the total. Of the remaining phrases, M phrases that have appeared in more document information are extracted as phrases representing the topic. In the phrase storage unit 205 of FIG. 5, the phrases “XX City”, “Satoshi Pinch”, “Pikachu's Special Move”, etc., stored in association with the program title “Anime C” are extracted in Step 405. This is an example of a phrase.

  Also, the phrase extraction method in step 409 differs from step 405 in that the phrase is extracted from one document information (subtitle or summary sentence), so the phrase used in more document information represents the topic. The phrase method is not suitable. Therefore, the following processing is performed.

  First, similarly to step 405, morphological analysis processing is performed to extract nouns, compound nouns, and noun phrases (= words). Since the summary sentence has the property that important information to be conveyed at the beginning of the sentence is described, a higher evaluation value A is given to the extracted word / phrase as it appears at the beginning. Moreover, since more accurate information can be conveyed as the number of characters increases, a higher evaluation value B is given as the number of characters increases. The word with the highest evaluation value is extracted from the evaluation values A and B. In FIG. 5, “ΔΔ and reunion” stored in association with the program title 901 “drama E” is an example of the phrase extracted in step 409.

  In the present invention, the operations shown in FIGS. 6, 7, 8, and 9 are constructed as a program and installed in a computer used as a phrase extracting device to be executed or distributed through a network. It is also possible.

  It is also possible to store the constructed program in a hard disk, a flexible disk, or a CD-ROM connected to a computer used as a phrase extraction device, and install and execute the program on the computer.

  The present invention is not limited to the above-described embodiments and examples, and various modifications and applications can be made within the scope of the claims.

  The present invention can be applied to a technique that uses a topical phrase according to the degree of attention of a program started at a specific time such as a television program or a radio program.

It is a figure for demonstrating the principle of this invention. It is a principle block diagram of this invention. It is a block diagram of the phrase extraction apparatus in one embodiment of this invention. It is a figure which shows the example of accumulation | storage of the program attention degree calculation and accumulation | storage part in one embodiment of this invention. It is a figure which shows the example of accumulation | storage of the phrase storage part in one embodiment of this invention. It is a flowchart of the process of the program information extraction part in one embodiment of this invention. It is a flowchart of the management process of the total result table according to program in one embodiment of this invention. It is a flowchart of a program attention degree calculation and process in one embodiment of the present invention. It is a flowchart of the process of the phrase extraction part in one embodiment of this invention. It is a figure which shows the process timing in one embodiment of this invention.

Explanation of symbols

100 Phrase Extraction Device 101 Program-Specific Document Extraction Unit, Program Information Extraction Unit 102 Program Attention Level Calculation / Means, Program Attention Level Calculation / Storage Unit 103 Program-Specific Document Information Storage Unit, Program-Specific Document Information Storage Unit 104 Phrase Extraction Unit, Phrase Extraction unit 201 Search term storage unit 202 Document information storage unit, document information storage unit 203 Program table storage unit, program information storage unit 204 Phrase storage unit, phrase storage unit 500 Total result table 501 by program Program ID
502 Program title 503 Cycle 504 Start date 505 End date 506 Attention calculation date 507 Total data 510 Program attention table 511 Rank 512 Program ID
513 Attention 901 Program title 902 Start date 903 Words

Claims (5)

Extract words used to extract and provide topics related to events that start at a specific time, such as TV programs and radio programs, from document information published on the network and program schedules that record program schedules A device ,
A program that searches and extracts a document including a program title from a set of documents having time information read from a document information storage unit that stores documents having time information, and stores the program in a document information storage unit for each program. A separate document extraction means ;
From the extracted documents , documents having time information within a predetermined attention level calculation period corresponding to the broadcast period of the program from the program start date and time are totaled for each program, and the program attention level is set as the program attention level. Attention level calculation means for each program stored in the table;
When the program attention level table is referenced, a program with a high degree of attention is selected in the previous broadcast, and the next broadcast of the program is within a preset time before the broadcast of the program until the broadcast start time Acquires the subtitle or outline of the program from the program guide storage means storing the program guide, performs morphological analysis, extracts words and phrases based on the appearance position and number of characters of the phrases, and starts broadcasting after the program broadcasts If it is within the preset time from the time, the document stored in the program-specific document information storage means is subjected to morphological analysis, the number of documents appearing for each phrase is obtained, and the phrases having a large number of appearing documents are extracted. A word extraction means to
The phrase extraction apparatus characterized by having a. The program attention level calculating means includes :
When the same user inputs the same keyword a plurality of times at short time intervals using the information specifying the user who has input the search word from the search words requested for the attention degree calculation period , once by counting, it counts the search terms occupancy of associated turn sets, according to claim 1 further comprising a means for storing the program of interest degree table the sum of the total value of the number of documents as a target of the program Phrase extraction device . The program attention level calculating means includes:
Claim 1 or phrase extraction apparatus 2 according to set a high degree of attention for the new program. The phrase extracting means is
The maximum period until the start of broadcasting is set in advance according to the program broadcasting cycle,
Seek time to broadcast begins at a start date and the current date and time of the program to be broadcast therefrom, with respect to program time to the broadcast start does not exceed the maximum period, according to claim 1 further comprising a means for phrase extraction phrase extraction device. A phrase extraction program for causing a computer to function as each means constituting the phrase extraction apparatus according to claim 1 .

Images