JPH0628416A - Multi-media information retrieving system - Google Patents

Abstract

PURPOSE:To provide the multi-media information retrieving system which retrieves information in a data, base and can provide the information of an expression method conforming to a user's request. CONSTITUTION:In an external information reference converting rule base 10, a converting rule group for showing the relation between a reference of a classification inputted by a user, and an attribute for expressiong the feature of information in a multi-media information store part 1, is stored. A necessary attribute deciding part 9 selects an attribute and a value required for retrieving the multi-media information store part 1 from the converting rule group by the reference of the classification and retrieval information of a keyword, etc., An external information reference converting part 12 retrieves the multi-media information store part 1 by using the selected attribute and value andcalculates the attribute and the value for showing the feature of the information in the obtained information, and weight of the information in the reference of the classification in accordance with the converting rule. A valid information deciding part 13 compares a threshold inputted by the user in advance and the weight of the information, and returns the information in which the weight of the information exceeds the threshold to the user.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention includes one or more pieces of information to be transmitted such as moving images, still images, voices, and texts, one or more attributes expressing the characteristics of the information, and values for the attributes. The present invention relates to a multimedia information retrieval system that retrieves a multimedia database stored together.

[0002]

2. Description of the Related Art In a conventional information retrieval system, particularly a multimedia information retrieval system for handling current information represented by newspaper articles and news manuscripts, retrieval is based on the date when a keyword or information was reported. It is difficult to select an appropriate keyword required for searching a database, and the following methods have been conventionally used to solve this problem.

Prepare a thesaurus dictionary corresponding to each keyword and search for more information.

A search condition written in natural language is input instead of a keyword, and an appropriate keyword is set in the natural language to perform a search (eg, Sugiyama et al. "Information retrieval system based on natural language understanding: IRIS", information IPSJ Natural Language Study Group Research Report Vol.58 No.8, 198
6). In these methods, the purpose of the user's search is comprehensive information search, and the search condition given by the user is
The assumption is that we cannot give enough keywords to perform an exhaustive search.

[0005]

However, for example, in the case of current affairs information represented as multimedia information, it is different from the purpose of exhaustive search which is performed by the conventional information retrieval database for text information. Since it has information from multiple media, it is becoming more and more objective to take a multifaceted view of one piece of information, regardless of the difference between senders of information such as newspapers and television stations.

In order to express multi-faceted information to the user, the criteria for classifying the information in the database by keywords and the like are not sufficient, and the criteria for the information classification required by the user and the user's request It is necessary to select and process the information necessary to match the expression method used.

[0007] As a result, the technique of generating a wide range of keyword groups from the search conditions based on the thesaurus or natural language to exhaustively search the database will blindly acquire a large amount of information, and the user will be blindly obtained. There was a drawback that the information could not be classified according to the purpose.

An object of the present invention is to provide a multimedia information retrieval system capable of retrieving information in a database and providing information of a representational method that meets a user's request.

[0009]

According to the multimedia information retrieval system of claim 1 of the present invention, in order to determine the importance of the information when displaying the retrieved information, the classification criteria inputted by the user and the multimedia are inputted. In order to search the multimedia database from the conversion rule group based on the conversion rule group showing the relationship between the attributes expressing the characteristics of information in the database and the search information such as the classification criteria and the keywords input by the user. Required attribute selection means for selecting required attributes and values, attributes and values representing information characteristics in information obtained as a result of searching the multimedia database using the selected attributes and values, and the conversion rule An information weight calculation means for calculating the weight of information in the classification criterion input by the user according to the conversion rule in the group, and a user in advance. The is by comparing the weight threshold and said information input, and a valid information determining conversion means to return as a result of the above information threshold system users.

In the multimedia information search system according to claim 2 of the present invention, the time when the information is announced is stored in the multimedia database as the information expressing the characteristic of the information, and the transmission time difference due to the medium through which the information is transmitted is calculated. The information weight calculation means has a time difference calculation means for calculating, and the information weight calculation means changes the weight of information according to the degree of possibility that information similar to the information that the user wants to retrieve exists in a certain time range.

In the multimedia information retrieval system according to claim 3 of the present invention, as the information expressing the characteristics of the information, there are information such as a place name, a person's name, a time when the information occurred, an object of the information, a cause of the information generated, and an operation. Is stored in the multimedia database, and has an occurrence frequency measuring means for measuring the occurrence frequency of each attribute and value, and the information weight calculating means, from the occurrence frequency of a certain attribute and value with respect to the time when the information occurs,
Calculate the weight of the information entered by the user on the basis of classification.

In the multimedia information retrieval system according to claim 4 of the present invention, as the information expressing the characteristics of the information, the country name where the information is reported, the place name, and the place name where the information is generated are stored in the multimedia database. An information frequency calculation means is used to measure the occurrence frequency of the combination of the country name or place name where the information was reported and the place name where the information is generated.The information weight calculation means is the classification criterion input by the user for the country name or place name. Calculate the weight of information in.

[0013]

According to the invention of claim 1, data for searching information such as a keyword inputted by a user and classification of information to be arranged from important information when displaying a result of the information search are classified. Select the attribute and value from the conversion rule group for searching the multimedia database using the standard data, and for the information obtained from the multimedia database, the attribute and value required for classification entered by the user. By calculating the weight of information when displaying information from the conversion rules in the conversion rule group, the difference between the user's information classification standard and the information classification standard in the database is resolved. And

For example, in television news and newspaper articles,
Newspaper articles are reported half a day later than TV news one day later, so TV news identical to a newspaper article should be searched for from news information about a half day back. Therefore, the invention of claim 2 stores the time when the information is transmitted in the multimedia database as an attribute representing the characteristic of the information, calculates the difference of the information transmission time due to the difference of the information transmission medium, and calculates the transmission time. By calculating the reliability of similar information (weight of information) from the difference between the two, it is possible to judge the similarity of information between different information of the information transmission medium.

For example, when expressing the history of the world as a user's classification standard, since history is caused by human behavior, it can be achieved by measuring the frequency of occurrence of country names and person names at a certain time. Therefore, the invention of claim 3 measures the occurrence frequency of the attribute in the multimedia database required for the classification attribute requested by the user at a certain time, and compares the attribute frequency in the multimedia database with respect to that time. It is characterized by calculating the weight of the information entered by the user based on the classification criteria using the magnitude of the frequency of occurrence, and determining the importance of the information according to the criteria requested by the user against changes over time. To do.

For example, when it is desired to create a world map that expresses the strength of relations between countries as a classification criterion of users, it depends on the country where certain information is generated, how the information is reported by the country that reports the information, and the frequency of occurrence. Can be calculated. Therefore, claim 4
Of the invention measures the frequency of occurrence of the country name and place name in which the information occurred, the country name reporting the information, and the rural area at a certain time, and the frequency of occurrence of the country name that the user wants to retrieve It is characterized in that (weight of information) is calculated, and the importance of the information inputted by the user with respect to the strength of the influence of the area is judged according to the classification standard.

According to the present invention, the time difference between the information due to the difference of the transmission medium is absorbed, and the television news having the same information as the newspaper can be accurately retrieved, or the information can be simply retrieved. Instead, in order to perform a search that matches the user's expression method, by inserting a rule for generating the user's expression method classification from the taxonomy in the multimedia database, for example, the importance of information between regions. Information such as changes in can be searched accurately.

[0018]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram of a multimedia information retrieval system according to an embodiment of the present invention.

In the multimedia information storage unit 1 which is a multimedia database, still images, moving images, audio,
Information such as text, attributes expressing the characteristics of those information, and their values are stored. Search information input section 2
Asks the user to input information for searching the multimedia information storage unit 1 and whether the information is a keyword,
Each type of input information (keyword, identifier) is determined by determining whether it is an identifier of the information that the information in the multimedia information storage unit 1 already searched for has, or a selection branch selected from the system, such as a menu. , Selector)
The input information is transmitted to the search information generating units 4, 5, 6 corresponding to. When the user inputs one or more keywords as information for performing a search, the search information generation unit 4 uses the input keywords to determine the attributes of the input keywords from the multimedia information storage unit 1. Extract and
It is expressed as group information based on keywords and attributes, and sent to the input information determination unit 7. The search information generation unit 5 stores multimedia information using the input identifier when the user inputs the identifier indicating the specific information obtained when the search is performed as the information for the user to perform the search. The section 1 is searched, all the sets of the value of the information having the input identifier and its attribute are extracted, and sent to the input information determining section 7. The search information generation unit 6 uses a menu provided by the system or direct input by the user as information for the user to search.
When the attribute and the value of the multimedia information storage unit 1 are input, the input data is converted into the set information of the value and the attribute that the medium is, for example, A newspaper, and is sent to the input information determination unit 7. The information obtained by the search information generators 4, 5 and 6 is composed of attributes and values that determine internal standards in the multimedia information storage unit 1. The input information determination unit 7 searches the multimedia information storage unit 1 by eliminating the duplication, because the information including the attributes and values generated by the retrieval information generation units 4, 5, and 6 may have duplication. Create a set of candidate attribute-value pairs required for. The attribute classification unit 8 classifies the set of candidates created by the input information determination unit 7 for each attribute according to the attribute attached to each candidate. The external information standard input unit 3 stores a standard of information classification (external information standard) input by the user when displaying the retrieved information. The external information standard conversion rule base 10 includes the knowledge of attributes and values necessary for searching the multimedia information storage unit 1 according to the external information standard input by the user, and the result of searching the multimedia information storage unit 1. , Conversion rules such as knowledge for calculating the weight of information according to the external information standard are stored.
The required attribute determination unit 9 which is a required attribute selection unit uses the conversion rules in the reference information conversion rule base 10 to classify the attributes and values required for searching the multimedia information storage unit 1 by the attribute classification unit 8. Select from among the specified attributes and values. The database search unit 11 searches the multimedia information storage unit 1 based on the attribute or value selected by the required attribute determination unit 9. The external information criterion conversion unit 12 which is an information weight calculation means extracts an attribute included in the external information criterion from the attribute obtained as a result of the database retrieval unit 11 retrieving the multimedia information storage unit 1 and a set of the values. The conversion rule of the external information standard conversion rule base 10 is used to convert the weight of information according to the external information standard. The valid information determining unit 13 which is a valid information determining unit compares the calculated weight of information with a threshold value given in advance, and the information obtained by searching the multimedia information storage unit 1 is finally obtained by the user. It is determined whether the information should be provided to the user.

Next, the processing steps in this embodiment will be described with reference to FIGS. 2 and 3 by taking a search of a database having newspapers and television news as multimedia information. 2 and 3 show a case where a keyword and an external information standard are input.

The user inputs one or more keywords and external information criteria for displaying the information requested by the user (step 20). The input keyword is identified as a keyword in the search information input unit 2,
It is sent to the search information generation unit 4 which generates search information from the keyword (step 21). Next, the search information generation unit 4 searches the multimedia information storage unit 1 using the input keyword group, and acquires all the attributes having the input keyword as a value (steps 22 to 24). Next, an attribute-value pair is generated from each acquired attribute and the corresponding value, and the attribute-value pair is transmitted to the attribute classification unit 8. The transmitted attribute-value pairs are classified by attribute and sent to the required attribute determination unit 9 as a value group for each attribute (step 25). In order to search the multimedia information storage unit 1, the required attribute determination unit 9 searches the external information criterion conversion rule base 10 using the external information criterion input by the user, and determines the classification attribute (required) for the external information criterion. A classification attribute is selected (step 26). Next, only the value group corresponding to the essential classification attribute is extracted from the value group classified by the attribute classification unit 8, and the attribute and the value group are searched for in the database search unit 11
Send to. (Steps 27-29). The database search unit 11 searches the multimedia information storage unit 1 using the given attribute and value group, and finds the corresponding attribute-value pair as 1.
Acquire all the information including one or more (step 30). Next, in the external information standard conversion unit 12, the external information standard conversion rule base 10 is searched by the external information standard in the external information standard input unit 3 to acquire the conversion rule (step 31). Next, the conversion rule obtained in step 31 is applied to all the information obtained in step 30, and the weight of information for each external information criterion for each information is calculated (steps 32 to 34). When the calculation of the information weight is completed, the information weight is compared with a threshold value given in advance for each external information standard, and only the information whose weight is equal to or greater than the threshold value set for all external information standards is compared. It is selected and returned to the user as a result (steps 35 to 37).

Next, this embodiment will be described with reference to a specific example.

The input information determining section 7 includes a search information generating section 4,
The information generated by 5, 6 is, for example, “keyword
Mayday, place Yoyogi Park, keyword Mayday,
In the case of a plurality of sets of information such as "organizational union", redundant "keyword mayday" is omitted and sent as "keyword mayday, place Yoyogi park, organization union" to the attribute classification unit 8.

The data in the multimedia information storage unit 1 is
As shown in Table 1, for each piece of information, one unique identifier, one or more attributes representing the characteristics of the information and their values (numerical values, keywords), still images used by the user,
It consists of three types of data sets of information itself such as moving images.
The attribute that represents the characteristics of information is the type of information used by the user such as still images, moving images, and audio, the sender of information such as a newspaper company or broadcasting station, one or more keywords that represent the characteristics of information, and the information that is transmitted. It consists of the time spent. As an example of concrete data, as shown in Table 1, image data of newspaper articles, information type “still image”, sender “A newspaper company”,
Keyword "Mayday", transmission time "May 2, 1992
It becomes the morning edition of "Sun" and the identifier is "1992252001".

[0026]

[Table 1] As shown in FIG. 4, the external information reference conversion unit 12 obtains from the information operation slot 42, the conversion rule obtained by searching the external information reference conversion rule base 10 by the external information reference, and the required attribute determination unit 9. A reference value for calculating the weight of information is created from the attribute and value obtained, and the information calculation slot 42
The slot initial setting unit 41 to be stored in, the value (reference value) of the information calculation slot 42, and the multimedia information storage unit 1
The information weight calculation unit 43, which is a transmission time difference calculation unit, which compares the information obtained from the above, determines the weight of the information, and stores it in the information calculation slot 42, and the value stored in each information calculation slot 42 (each slot). Is set by the information weight calculation unit 43 in accordance with the value of the external information reference, and the slot value calculation unit 44 that calculates the importance (information weight) of the information expressed by each slot. Further, the information output control unit 4 for determining whether to output the weight of information attached to each piece of information as a result or output the weight of information in the slot calculated by the slot value calculation unit 44.
It consists of 5.

The conversion rule of the standard information conversion rule base 10 includes, for example, a condition based on one or more external information standards, and an attribute type required when searching the multimedia information storage unit 1 to satisfy the condition. Depending on the case, it consists of a calculation formula for calculating the weight of information using the value and the attribute. As a concrete example of the conversion rule, for example, the conversion rule corresponding to “the content is the same and TV news” is, as the attribute type, the keyword and the sender are ATV.
As shown in the procedure of Fig. 5, each TV station has a broadcasting station such as, and the same information is reported in TV news and newspaper articles, but a delay of half a day to a day is reported in newspaper articles. Therefore, in order to compare the time obtained from the information input by the user with the result obtained from the multimedia information storage unit 1, in the slot initial setting unit 41, if the information input by the user is a morning article, , Candidate time 1
From the afternoon of the previous day to 1:00 am of the day, the weight is 1.
A reference value of 0, a candidate time of 2 from midnight to noon of the previous day, a weight of 0.8, a candidate time of 3 from 0.5 to the day before yesterday, a candidate time of 4 from 2 days ago, and a weight of 0.0 today is generated. (Step 51). Then, the value of each candidate time and the weight of information are stored in the information calculation slot 42 (step 52). Next, in the information weight calculation unit 43, for each piece of information obtained as a result of searching the multimedia information storage unit 1, the information calculation slot 42 is searched using the time that each piece of information has, and the weight of the information is obtained. Then, according to the external information standard, the information output control unit 45 sends the output (information weight) from the information weight calculation unit 43 to the valid information determination unit 13 (steps 53 to 55).

The processing of the necessary information determination unit 9, the external information standard conversion rule base 10, the database search unit 11, and the external information standard conversion unit 12 is performed by a multimedia database having image data of newspapers and moving image data of TV news. This will be explained as an example. The input from the user is, for example, the identifier 199252001 of the article about Mayday, which was already searched on May 2, 1992, and the attribute input by the user that expresses the searched information is the same information but information from different media. Suppose there is. The input to the required attribute determination unit 9 is, for example, “transmission time May 2, 1992 morning edition, sender A newspaper publisher, keyword Mayday”. In the necessary attribute determination unit 9, the conversion rule matching the condition “contents are the same and TV news” obtained from the external information standard input unit 3 is converted to the external information standard conversion rule base 1
Get from 0. For example, as the type of attribute, "keyword, sender ATV, sender BTV, sender CTV, sender DT
The information for searching the multimedia information storage unit 1 such as "V" is obtained. Although the value of the keyword cannot be obtained from the external information standard conversion rule base 10, the keyword becomes “Mayday” due to the data from the attribute classification unit 8. Therefore, in the database search unit 11, "Keyword May Day,
Caller ATV, Caller BTV, Caller CTV, Caller D
Search by "TV". As a result, for example, "caller ATV, call time May 1, 1992 7:00 pm, identifier 199251702", "caller ATV, call time 1
May 1, 991 7:00 pm, identifier 19991570
2 ”,“ sender CTV, transmission time 6:00 pm May 1, 1992, identifier 199211003 ”can be obtained. The external information criterion conversion unit 12 determines the weight of information for each piece of information acquired by the database search unit 11 by the procedure shown in FIG.
As a result, for example, "sender ATV, transmission time May 1, 1992 7:00 pm, identifier 199251702" and "sender CTV, transmission time May 1, 1992 6:00 pm, identifier 1992511003" have information weights. 1.0, the weight of the information is 0. "Sender ATV, transmission time May 1, 1991 7:00 pm, identifier 199151702".
It becomes 0. In the valid information determination unit 13, the weight of the information determined by the external information reference conversion unit 12 is compared with a threshold value included in the external information reference input by the user in advance, and when the threshold value is 1.0, for example. , "Caller AT
V, calling time May 1, 1992 7:00 pm, identifier 19
9251702 "and" Caller CTV, call time 1992
On May 1, 6:00 pm, the information of the identifier 1992511003 "is returned to the user.

There is a case where the information of the newspaper is acquired in order to obtain a more detailed explanation while watching the television news, but by reversing the information of the time difference between the newspaper and the television news in this embodiment, this information can be obtained. The description of the case is omitted because it is easily analogized.

Next, an embodiment of the invention described in claim 3 will be described in detail with reference to the drawings.

The system configuration in this embodiment is the same as in FIG. As shown in Table 2, the data of the multimedia information storage unit 1 in the present embodiment includes, for each information, one unique identifier, one or more attributes and values indicating the characteristics of the information, and a user. It consists of three types of data sets of information itself such as still images and moving images used by. The attributes and values representing the characteristics of information in this embodiment include, for example, in addition to the attributes described in the first embodiment, the country name (Japan) where the information occurred, the city name (Tokyo), the name of the person who generated the information ( Prime Minister Miyazawa), organization name (Liberal Democratic Party), and information target (PK
O bill).

[0032]

[Table 2] The attributes that represent the characteristics of other information in this embodiment are:
5W1H often used to express current affairs information
(When, Where, Who, What, Why,
How) classification methods are also possible, for example When “19
May 25, 1992 ", Where" Japan, Tokyo ", Wh
o "Prime Minister Miyazawa", What "correction of the lower house number", Why
"One vote disparity", How "diet deliberation" is also possible.

As the conversion rule of the external information standard conversion rule base 10 in the present embodiment, for example, the conversion rule "from year to year and the effect between countries" is as follows:
As shown in FIG. 6, there are two country names as the attributes of the database search, and the procedure is as shown in FIG. 6, where the weight of the information of each field is the slot initial in accordance with the time step input as the external information standard and the field that the information represents. From the setting section 41,
It is set in the information calculation slot 42 which is a generation frequency measuring means (step 61). Next, in the information weight calculation unit 43, the reference value of the information calculation slot 42 corresponding to the time step and the field of each information is incremented by 1 with respect to the result obtained from the multimedia information storage unit 1 ( Steps 63, 64). Next, the slot value calculation unit 44 calculates the importance at each time step according to the conversion rule. For example, in the case of an influence between nations, this processing is roughly according to types such as politics, economy, and culture, more precisely, car trade and athletes. For example, the external information standard of the user is "impact due to US friction". If so, the weight of economic and political fields is set to 1.0, the weight of science and technology is set to 0.5, and the weight of culture is set to 0.2. The strength is expressed by the sum of the product of the value of each attribute and the weight of information. next,
The information output unit 45 sends the data from the slot value calculation unit 44 to the valid information determination unit 13 according to the external information standard input by the user (step 68).

In the present embodiment, the required attribute determination unit 9, the external information standard conversion rule base 10, the database search unit 1
1. The processing of the external information standard conversion unit 12 will be described as an example of searching for “a period of strong influence from 1960 to 1990 in Japan and the United States”. The input from the user is, for example, “Japan, United States, 1960-19
"90 years", and as the external information standard, "from year to year, influence between countries and between countries" is input. The input to the necessary attribute determination unit 9 is, for example, “calling time January 1, 1960 to December 31, 1990, target country Japan, United States”. In the necessary attribute determination unit 9, the conversion rule based on the conversion rule based on the external information standard which is obtained from the external information standard input unit 3 and which meets the condition of “the influence between the countries in the year from year to year” is obtained. Get from 10. As a result, for example, information for searching the Amachimedia information storage unit 1 such as "sender A newspaper, B newspaper, ATV, sender BTV, sender CTV, sender DTV, target country, time of origin" as the attribute type. obtain. Although the value of the target country and the transmission time cannot be obtained from the external information standard conversion rule base 10, the attribute classification unit 8
According to the data from, the target countries are “Japan” and “United States”, and the transmission time is “January 1, 1960 to December 31, 1990”. Therefore, the database search unit 1
In 1, "Sender A newspaper, B newspaper, ATV, sender BT
V, caller CTV, caller DTV, target country Japan, United States, call time ". As a result, for example, "sender ATV, transmission time September 1, 1990
7:00 pm on Sunday, identifier 199082901, field politics "," Sender A newspaper, dispatch time morning, August 29, 1990, identifier 1990831002, sector politics, "sender A newspaper, dispatch time September 1, 1990 morning publication , Identifier 199091911, field economy ”can be obtained. In the external information reference conversion unit 12, for each piece of information acquired by the database search unit 11, with respect to the time step determined by the procedure shown in FIG.
Calculate the impact strength. First, the slot initial setting section 41
From 1960 to 1990 in 1 year increments,
And the slot is initialized for each field (step 6).
1). The initial value is 0. Database search unit 11
The information weight calculation unit 43 increases the value slot of the reference value corresponding to the information by 1 from the information weight calculation unit 43 according to the attribute and value of the information obtained from Send instructions (steps 62-6)
4). When all the data are processed, the information calculation slot 42 sends the value for each year and the weight of the information for each field (FIG. 3) to the slot value calculation unit 44, and multiplies the value and the weight of the information for each field. , The sum is calculated and used as the weight of information (steps 65 to 67). As a result, if the preset time step is one year, 19
The weight of the information for 90 years is about 1500 from the weight and the value of the information in Table 3 when only the newspapers A and ATV are made into a database. Next, the weight of information is sent to the valid information judging section 13 (step 68). In the valid information determination unit 13, the user previously inputs the weight of the information determined by the external information reference conversion unit 12,
Compared with the threshold included in the attribute expressing information, for example, when the threshold is 1000, 1960 to 199
All years up to year 0 are judged to have a large impact.

On the other hand, if the same processing is performed for Japan and Iraq, for example, it will be about 60 in 1989, but will be about 1400 in 1990, and the data of 1990 will be larger than the threshold value of 1000. It is determined that the relationship between India and Iraq has become stronger.

As an example of data in the information calculation slot 42, as shown in Table 3, the year in which the information is generated, the field indicating the attribute, the reference value consisting of the weight of each information with respect to the external information reference, and the reference value are shown. It consists of a value that represents the number of times the reference was made.

[0037]

[Table 3] When the slot initial setting unit 41 performs a search including 1992 on June 1, 1992, the slot initial setting unit 41 displays 199.
Since there is only about 150 days of information in 2 years, the value in the slot is absolutely smaller than in other years. Therefore, regarding the time step including the time when the search is performed, the weight of the information can be normalized by setting the weight of the information to be the number of days of the time step / the number of days until the search day as compared with other time steps. For example, if you perform a search on June 1, 1992, and if the time step is one year, it is not a leap year,
2.4 times.

Next, an embodiment of the invention of claim 4 will be proved.

In this embodiment, the information of the country name or the place name where the information is reported is added to the data structure of the second embodiment. External information criterion conversion rule base 1 in this embodiment
As a conversion rule of 0, for example, there is a conversion rule of “from year to year, worldwide distribution of influence between countries”.

In the database search section 11, the "sender A
Newspaper, B newspaper, ATV, sender BTV, sender CTV,
Search by "Caller DTV, target country Japan, United States, calling time, calling country Japan, calling country China, calling country United States, calling country France, calling country Egypt Republic, calling country Thailand". As a result, for example, "the originating country Japan, the originating time September 1, 1990
7:00 pm on Sunday, identifier 19908289001, field politics "," Source country Japan, transmission time August 29, 1990 morning edition, identifier 1990831002, field politics "," Source country Japan, transmission time September 1, 1990 morning edition, identifier 19
Information such as "9099111, sectoral economy" can be obtained.

In the external information reference conversion section 12, for each piece of information acquired by the database search section 11, for each time step determined by the procedure shown in FIG.
Calculate the impact strength. First, the slot initial setting section 41
From 1960 to 1990 in 1 year increments,
At the same time, the slots are initialized for each field and each source country (step 61). The initial value is 0. With respect to the information obtained from the database search unit 11, the information weight calculation unit 43 sends an instruction to the information calculation slot 42 to increase the value corresponding to the information by 1 (steps 62 to 64). After processing all information, from the information calculation slot 42,
The slot value calculation unit 4 calculates the value for each year and the weight for each field.
Send to 4 and multiply by the value and the weight for each field to calculate the sum,
The weight of information is used (steps 65 to 67). As a result,
If the preset time interval is one year, the weight of the information is, for example, for Japan-US relations, the originating country is 3000 in Japan, the originating country in France is 150, and the United States is 1000. It is possible to judge the difference in the interests of each country.

In the data structure of the information calculation slot 42 in this embodiment, as shown in Table 4, the country of origin of information is added to the attributes of Table 3.

As one embodiment of the slot value calculation unit 44, the total number of information for each source country in the multimedia information storage unit 1 is calculated for the calculated weight of each information, and the information for each source country is calculated. By normalizing the weight of information according to the ratio of the amount of information transmitted, the importance of certain information for that country can be expressed more accurately.

[0044]

[Table 4]

[0045]

As described above, the present invention has the following effects. (1) The invention of claim 1 selects search information such as a keyword, which is input by a user, and an attribute and a value for searching a database using data of a criterion for classifying the information from the conversion rule group, By calculating the weight of information in the classification criterion input by the user using the attribute and value and returning the information having the information weight of the threshold value or more to the user, the criterion of the information classification of the user and the information in the database The difference with the information classification standard can be eliminated. (2) According to the invention of claim 2, the time when the information is transmitted is stored in the database, the difference in the transmission time of the information due to the difference of the information medium is calculated, and the weight of the information is changed from the difference. , It is possible to judge the similarity of information between different information transmission media. (3) According to the invention of claim 3, the frequency of occurrence of each attribute and value is measured, and the weight of the information input by the user by the classification standard is calculated using the frequency of occurrence, whereby It is possible to judge the importance of information on the criteria required by all users. (4) The invention of claim 4 calculates the frequency of occurrence of information or the country or place name that is generating information, and calculates the weight of the information based on the classification entered by the user based on the frequency of occurrence. By doing so, it is possible to determine the importance of the criterion requested by the user with respect to the strength of the influence of the area.

As described above, according to the present invention, when the user displays information in advance, the difference between the temporal and attribute classification criteria and the temporal and spatial classification of the database system is absorbed. Information is added as a database, and a mechanism to calculate the weight of information using the information in the database is provided in accordance with the classification criteria when expressing the information of the user. On the other hand, it is possible to provide a multi-faceted method of expressing information.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a multimedia information search system according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a process of the multimedia information search system of FIG.

FIG. 3 is a flowchart showing a process of the multimedia information search system of FIG.

FIG. 4 is a configuration diagram of an external information reference conversion unit 12 in FIG.

FIG. 5 is a flowchart showing a processing procedure of an external information reference conversion unit 12.

FIG. 6 is a flowchart showing another processing procedure of the external information reference conversion unit 12.

[Explanation of symbols]

 1 multimedia information storage unit 2 search information input unit 3 external information reference input unit 4 search information generation unit 5 search information generation unit 6 search information generation unit 7 input information determination unit 8 attribute classification unit 9 required attribute determination unit 10 external information standard Conversion rule base 11 Database search unit 12 External information reference conversion unit 13 Effective information determination unit 20 to 37 Step 41 Slot initial setting unit 42 Information calculation slot 43 Information weight calculation unit 44 Slot value calculation unit 45 Information output control unit 51 to 55, 61-68 steps

Claims (4)

[Claims] 1. A multimedia in which information to be transmitted such as one or more moving images, still images, voices, and texts, one or more attributes expressing the characteristics of the information, and a value for the attribute are stored together. In a multimedia information retrieval system for searching a database, in order to determine the importance of the information when displaying the retrieved information, the classification criteria entered by the user and the attributes expressing the characteristics of the information in the multimedia database. A conversion rule group representing the relationship between the conversion rule group and attributes and values required for searching the multimedia database from the conversion rule group are selected based on the classification information entered by the user and search information such as keywords. It was obtained as a result of searching the multimedia database using the required attribute selection means and the selected attributes and values. Information weight calculation means for calculating the weight of information in the classification criterion input by the user according to the attribute and value indicating the information feature in the information and the conversion rule in the conversion rule group; A multimedia information search system, comprising: effective information determining means for comparing information weights and returning information having a weight of information equal to or more than a threshold value to a user as a result. 2. As information representing the characteristics of the information, the time when the information is announced is stored in the multimedia database, and the transmission time difference calculation means for calculating the transmission time difference depending on the medium through which the information is transmitted is provided. The multimedia information search system according to claim 1, wherein the information weight calculation means changes the weight of information according to the degree of possibility that information similar to the information that the user wants to search exists within a certain time range. 3. A place name, which is information representing the characteristics of the information,
Information such as a person's name, time when information is generated, information object, cause of information generation, and behavior is stored in the multimedia database, and has an occurrence frequency measuring means for measuring the occurrence frequency of each attribute and value, 2. The multimedia information retrieval system according to claim 1, wherein the information weight calculation means calculates the weight of the information input by the user on the basis of the classification based on the occurrence frequency of a certain attribute and value with respect to the time when the information occurred. 4. As the information expressing the characteristics of the information, the name of the country or place where the information is reported and the place name where the information is generated are stored in the multimedia database, and the name of the country or place where the information is reported and the information are stored. There is an occurrence frequency measuring means for measuring the occurrence frequency of the combination with the generated place name, the information weight calculating means, the user input for the country name or place name,
The multimedia information retrieval system according to claim 1, wherein a weight of information on a classification basis is calculated.