System and method for determining credibility of information based on many remarks on a network, and non-transitory computer readable storage medium having stored thereon computer program therefor
A credibility determining system enabling highly accurate credibility determination of given information includes causal relation knowledge DB 90 and search scope constraint DB 88 . Causal relation knowledge each includes a combination of cause and result parts, and time and position constraints associated with the causal relation. The credibility determining system further includes a query generating unit 92 retrieving a causal relation matching the input information and based on the time and position constraints stored in search scope constraint DB 88 in association with the retrieved causal relation, generating a query for retrieving remarks from mini-blog text DB 84 , a text search unit 96 searching for related remarks from mini-blog text DB 84 using the query, and a display candidate selecting unit 100 classifying the searched remarks to expressions interpreted as causes or to results, determining credibility of the input information based on the classification result and outputting the determination result.
1. A computer-implemented credibility determining system, for determining credibility of input information based on a set of remarks on a network, the credibility determining system comprising:
a network interface connected to the network; and
a non-transitory computer readable storage medium; wherein the non-transitory computer readable storage medium includes:
a causal relation storage device for storing pieces of causal relation knowledge, each piece of the causal relation knowledge including:
a combination of a cause part consisting of a text representing a cause and a result part consisting of a text representing a result of the cause, and
a time constraint and a position constraint associated with the causal relation;
the non-transitory computer readable storage medium further includes a remark text storage device for storing a set of texts of the remarks;
the credibility determining system further comprising a processor connected to the network interface and the storage medium, wherein:
the processor is configured to perform:
searching for and retrieving, from the causal relations stored in the causal relation storage device, a causal relation matching the input information;
specifying, for each retrieved causal relation, a constraint condition on retrieval of a remark from the set of texts of the remarks based on a time constraint and a position constraint stored in the causal relation storage device in association with the causal relation;
for each retrieved causal relation, retrieving, from the set of texts of the remarks, a remark satisfying the specified constraint condition and representing a cause or a result of an item represented by the input information, using the causal relation knowledge;
classifying each of the retrieved remarks to those remarks representing causes of the item represented by the input information and to those remarks representing results of the item represented by the input information, determining credibility of the input information based on the classification, and outputting the result of determination; and
extracting, from the set of texts of the remarks, for each retrieved causal relation, a remark representing a cause or a result of an item represented by the input information, which has a content satisfying the time constraint with the time of occurrence estimated for the input information being a point of reference or a content satisfying the position constraint with the position of occurrence estimated for the input information being a point of reference,
wherein in order to extract the remark, the processor is configured to perform:
specifying a first causal relation having in the result part thereof the input information or a text equivalent to the input information, from the causal relations stored in the causal relation storage device, and
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to the specified first causal relation and satisfying the time constraint and the position constraint, respectively, associated with the specified first causal relation, with the estimated time and the estimated position being points of references.
2. The credibility determining system according to claim 1 , wherein the processor is further configured to perform repeatedly collecting remarks from the network through the network interface and storing them in the remark text storage device.
3. The credibility determining system according to claim 2 , wherein
each of the constraint conditions includes a time constraint condition or a geographical constraint condition to be established between a cause and a result of the causal relation; and
in order to realize retrieval from the set of texts of the remarks, the processor is configured to perform:
estimating a time of occurrence of an item represented by the input information, based on time information obtained from the input information, and
estimating a position of occurrence of an item represented by the input information based on geographical information obtained from the input information.
4. The credibility determining system according to claim 3 , wherein
in order to extract the remark, the processor is further configured to perform:
specifying a second causal relation having in the cause part thereof the input information or a text equivalent to the input information, from the causal relations stored in the causal relation storage device, and
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to the result part of the specified second causal relation and satisfying the time constraint and the position constraint, respectively, stored in the causal relation storage device associated with the specified second causal relation, with the estimated time and the estimated position being points of references.
5. The credibility determining system according to claim 4 , wherein
the processor is further configured to perform extracting, from the set of texts of the remarks, for each the specified first or second causal relation, a remark contradicting a cause or a result of an item represented by the input information and satisfying the time constraint with the time of occurrence estimated for the input information being a point of reference and satisfying the position constraint with the position of occurrence estimated for the input information being a point of reference;
in order to perform the determination, the processor is configured to perform:
classifying each of the extracted remarks to a first type of remarks representing a cause of the item represented by the input information, to a second type of remarks representing a result of the item represented by the input information, and to a third type of remarks representing contradictions with the item represented by the input information,
determining credibility of the input information as a function of the number of remarks classified to the first type, the number of remarks classified to the second type and the number of remarks classified to the third type, by the classification, and
outputting the determined credibility; and wherein
a value of the function changes with positive correlations with respect to the numbers of remarks classified to the first and second types, and changes with a negative correlation with the number of remarks classified to the third type.
6. The credibility determining system according to claim 4 , wherein
in order to perform the determination, the processor is configured to perform:
classifying each of the extracted remarks to a first type of remarks representing a cause of the item represented by the input information and a second type of remarks representing a result of the item represented by the input information,
determining credibility of the input information as a function of the number of remarks classified to the first type and the number of remarks classified to the second type, and
outputting the determined credibility.
7. The credibility determining system according to claim 6 , wherein the function has a positive correlation with the number of remarks classified to the first type and the number of remarks classified to the second type.
8. The credibility determining system according to claim 7 , wherein an increment of a value of the function per unit of change of the number of remarks classified to the first type is larger than an increment per unit of change of the number of remarks classified to the second type.
9. A computer-implemented credibility determining method for determining credibility of input information based on a set of texts of remarks on a network, wherein
the computer includes:
a network interface connected to the network, and
a non-transitory computer readable storage medium;
the non-transitory computer readable storage medium includes a causal relation storage device for storing pieces of causal relation knowledge, each piece of the causal relation knowledge including:
a combination of a cause part consisting of a text representing a cause and a result part consisting of a text representing a result of the cause, and
a time constraint and position constraint associated with the causal relation;
the non-transitory computer readable storage medium further includes a remark text storage device for storing a set of texts of the remarks;
the credibility determining method comprising the steps of:
searching for and retrieving, from the causal relations stored in the causal relation storage device, a causal relation matching the input information,
for each retrieved causal relation, specifying a constraint condition for retrieving a remark from the set of texts of the remarks based on a time constraint and a position constraint stored in the causal relation storage device in association with the causal relation;
for each retrieved causal relation, retrieving, from the set of texts of the remarks, a remark satisfying the specified constraint condition and representing a cause or a result of an item represented by the input information, using the causal relation; and
classifying each of the retrieved remarks to remarks representing causes of the item represented by the input information and to remarks representing results of the item represented by the input information, determining credibility of the input information based on results of classification, and outputting the result of determination,
wherein the step of retrieving from the set of texts of the remarks includes the steps of:
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to a causal part of the specified first causal relation and satisfying the time constraint and the position constraint, respectively, associated with the specified first causal relation, with the estimated time and the estimated position being points of references,
specifying a second causal relation having in the cause part thereof the input information or a text equivalent to the input information, from the causal relations stored in the causal relation storage device, and
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to a result part of the specified second causal relation and satisfying the time constraint and the position constraint, respectively, associated with the specified second causal relation, with the estimated time and the estimated position being points of references.
10. The credibility determining method according to claim 9 , further comprising the step of repeatedly collecting remarks from the network through the network interface and storing collected remarks in the remark text storage device.
11. The credibility determining method according to claim 10 , wherein
each of the constraint conditions includes a time constraint condition or a geographical constraint condition to be established between a cause and a result of a causal relation; and
the step of retrieving from the set of texts of the remarks includes the steps of:
estimating a time of occurrence of an item represented by the input information, based on time information obtained from the input information;
estimating a position of occurrence of an item represented by the input information based on geographical information obtained from the input information; and
extracting, from the set of texts of the remarks, for each retrieved causal relation, a remark representing a cause or a result of an item represented by the input information, which has a content satisfying the time constraint with the time of occurrence estimated for the input information being a point of reference or a content satisfying the position constraint with the position of occurrence estimated for the input information being a point of reference.
12. The credibility determining method according to claim 11 , wherein
the step of extracting a remark includes the steps of:
specifying a first causal relation having in the result part thereof the input information or a text equivalent to the input information, from the causal relations stored in the causal relation storage device.
13. The credibility determining method according to claim 12 , further comprising the step of
extracting, from the set of texts of the remarks, for each specified first or second causal relation, a remark contradicting a cause or a result of an item represented by the input information and satisfying the time constraint with the time of occurrence estimated for the input information being a point of reference and satisfying the position constraint with the position of occurrence estimated for the input information being a point of reference; wherein
the determining step includes the steps of:
classifying each of the extracted remarks to a first type of remarks representing a cause of the item represented by the input information, a second type of remarks representing a result of the item represented by the input information, or a third type of remarks representing contradictions with the item represented by the input information,
determining credibility of the input information as a function of the number of remarks classified to the first type, the number of remarks classified to the second type and the number of remarks classified to the third type, by the classification, and
outputting the determined credibility; and wherein
a value of the function changes with positive correlations with respect to the number of remarks classified to the first type and the number of remarks classified to the second type, and changes with a negative correlation with the number of remarks classified to the third type.
14. The credibility determining method according to claim 12 , wherein
the determining step includes the steps of:
classifying each of the extracted remarks to a first type of remarks representing a cause of the item represented by the input information or a second type of remarks representing a result of the item represented by the input information,
determining credibility of the input information as a function of the number of remarks classified to the first type and the number of remarks classified to the second type, and
outputting the determined credibility.
15. The credibility determining method according to claim 14 , wherein the function has a positive correlation with the number of remarks classified to the first type and the number of remarks classified to the second type.
16. The credibility determining method according to claim 15 , wherein an increment of a value of the function per unit of change of the number of remarks classified to the first type is larger than an increment per unit of change of the number of remarks classified to the second type.
17. A non-transitory computer-readable storage medium having stored thereon a computer program for controlling a computer such that, based on a set of remarks on a network, the computer determines credibility of input information, wherein:
the computer includes:
a network interface connected to the network, and
the non-transitory computer readable storage medium;
the non-transitory computer readable storage medium includes a causal relation storage device for storing pieces of causal relation knowledge, each piece of the causal relation knowledge including a combination of a cause part consisting of a text representing a cause and a result part consisting of a text representing a result of the cause, and a time constraint and a position constraint associated with the causal relation;
the non-transitory computer readable storage medium further includes a remark text storage device for storing a set of texts of the remarks;
the computer program is configured to cause the computer to perform the steps of:
searching for and retrieving, from the causal relations stored in the causal relation storage device, a causal relation matching the input information, and for each retrieved causal relation, specifying a constraint condition for retrieving a remark from the set of texts of the remarks based on a time constraint and a position constraint stored in the causal relation storage device in association with the causal relation;
for each retrieved causal relation, retrieving, from the set of texts of the remarks, a remark satisfying the specified constraint condition and representing a cause or a result of an item represented by the input information, using the causal relation knowledge; and
classifying each of the retrieved remarks to those remarks representing causes of the item represented by the input information or to those remarks representing results of the item represented by the input information, determining credibility of input information based on results of classification, and outputting the result of determination from the computer,
wherein the step of retrieving from the set of texts of the remarks includes the steps of:
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to a causal part of the specified first causal relation and satisfying the time constraint and the position constraint, respectively, associated with the specified first causal relation, with the estimated time and the estimated position being points of references,
specifying a second causal relation having in the cause part thereof the input information or a text equivalent to the input information, from the causal relations stored in the causal relation storage device, and
extracting, from the set of texts of the remarks, a remark having a content matching or equivalent to a result part of the specified second causal relation and satisfying the time constraint and the position constraint, respectively, associated with the specified second causal relation, with the estimated time and the estimated position being points of references.
18. The storage medium according to claim 17 , wherein
each of the constraint conditions includes a time constraint condition or a geographical constraint condition to be established between a cause and a result of the causal relation; and
the step of retrieving includes the steps of:
estimating a time of occurrence of an item represented by the input information, based on time information obtained from the input information;
estimating a position of occurrence of an item represented by the input information based on geographical information obtained from the input information; and
extracting, from the set of texts of the remarks, for each retrieved causal relation, a remark representing a cause or a result of an item represented by the input information, which has a content satisfying the time constraint with the time of occurrence estimated for the input information being a point of reference or a content satisfying the position constraint with the position of occurrence estimated for the input information being a point of reference.