Patent Application
20100312792
The present invention relates to an information analysis (analyzing) device, an information analysis (analyzing) method, an information analysis (analyzing) program and a search system for evaluating the credibility of information.
In recent years, a large amount of information has been circulating throughout the world via Web pages, electronic bulletin boards and the like on the Internet. Accordingly, it is becoming more difficult for an ordinary user to determine which pieces of information are reliable or not on the Internet.
For example, when one proposition, such as “green tea works well for cancer” or “Pluto is a planet,” is picked up, a large amount of information about arguments for and against the proposition, including various articles, comments and opinions about the proposition, is circulating on the Internet. Therefore, a user can judge the credibility of the proposition by referring to a wide range of such information. However, the judgment may be biased by referring to only a part of the information. Meanwhile, in reality, it is extremely difficult for a user to go through all the information. In particular, the trend grows stronger as the credibility of the information changes with time.
For example, the credibility of the proposition that “Pluto is a planet” was changed in 2006, when the definition of what a planet is was changed. There is risk that the user's judgment may vary depending on when the information that the user refers to is posted.
The related technique for evaluating the credibility of information on the Internet is, for example, disclosed in NPL 1. According to the related technique disclosed in NPL 1, a large amount of Web documents including a specific proposition is classified by sender of document, approval or disapproval opinion of document, and meaning such as factual grounding before being presented, thereby helping a user judge the credibility concerning the proposition.
In many cases, time information, such as a date and time of creation or a date or time of sending, is added to articles, blogs, emails and the like on the Internet. There is a technique of extracting from information on the Internet documents including the proposition on which attention is focused, sorting the extracted documents by time information attached to each document, and presenting the time when the proposition on which attention is focused appears and the number of times that the proposition appears. For example, what is disclosed in NPL 2 is a method of extracting and displaying the time-series changes in the number of times that a specific topic word appears in the whole of collected blogs.
{NPL 1} H. Miyamori, et. al., “Evaluation Data and Prototype System WISDOM for Information Credibility Analysis”, In Proc. of First International Symposium on Universal Communication, pp. 234-237 (2007)
{NPL 2} Tomoyuki Nanno, Yasuhiro Suzuki, Toshiaki Fujiki, Manabu Okumura “Automatically Collecting and Monitoring Blogs,” Transactions of the Japanese Society for Artificial Intelligence, Vol. 19, No. 6, pp. 511-520, 2004
However, the problem is that even if the related technique disclosed in NPL 1 or NPL2 is used, it is impossible to appropriately evaluate the credibility of linguistic expressions to be analyzed at a specific point in time. The reason is that according to the related technique disclosed in NPL 1, the changes of credibility over time are basically not taken into account. Even when the related technique disclosed in NPL 2 of handling the changes over time is used, it is only possible to present the time-series changes concerning the number of times that a linguistic expression as an analyzation object appears. The time-series changes, such as increases or decreases in the number of times that the linguistic expression appears, does not necessarily reflect credibility.
For example, the following looks at the situation where a downward trend in the number of times that a linguistic expression corresponding to a given proposition appears is observed. In this case, even if the related technique disclosed in NPL 1 or NPL 2 is used, it is not possible to judge that the number of times that the linguistic expression appears is decreasing because of the denial of the proposition or that the proposition is less discussed because the proposition turns out to be a fact generally accepted by society.
There is a certain kind of proposition that turns out to be true only under a specific condition. However, it is impossible to capture the phenomenon of the proposition turning out to be true only under the specific condition just by observing the number of times that a linguistic expression as an analyzation object appears.
The exemplary object of the present invention is to provide an information analysis device, an information analysis method and an information analysis program that can appropriately evaluate the credibility of a linguistic expression as an analyzation object at a specific point in time.
According to the present invention, an exemplary information analysis device includes: a related expression generation section that accepts a evaluation object expression as input, the evaluation object expression being an linguistic expression to be evaluated, and generates a linguistic expression related to the evaluation object expression as a related expression; and a credibility calculation section that acquires the evaluation object expression and the related expression from a plurality of electronic documents along with time information and calculates credibility concerning the meaning of the evaluation object expression at a specific point in time by comparing the number of times that the acquired evaluation object expression appears and the number of times that the acquired related expression appears at the same time point.
According to the present invention, an exemplary information analysis method includes: accepting an evaluation object expression as input, the evaluation object expression being an linguistic expression to be evaluated, and generating a linguistic expression related to the evaluation object expression as a related expression; and acquiring the evaluation object expression and the related expression from a plurality of electronic documents along with time information and calculating credibility concerning the meaning of the evaluation object expression at a specific point in time by comparing the number of times that the acquired evaluation object expression appears and the number of times that the acquired related expression appears at the same time pointduring the same period.
According to the present invention, an exemplary information analysis program causes a computer to execute: a related expression generation process of accepting a evaluation object expression as input, the evaluation object expression being an linguistic expression to be evaluated, and generating a linguistic expression related to the evaluation object expression as a related expression; and a credibility calculation process of acquiring the evaluation object expression and the related expression from a plurality of electronic documents along with time information and calculating credibility concerning the meaning of the evaluation object expression at a specific point in time by comparing the number of times that the acquired evaluation object expression appears and the number of times that the acquired related expression appears at the same time point.
According to the present invention, it is possible to appropriately evaluate the credibility of an evaluation object linguistic expression at a specific point in time.
{FIG. 1} A block diagram illustrating an example of an information analysis device according to the present invention.
{FIG. 2} A flowchart illustrating an example of a process for the information analysis device to analyze information.
{FIG. 3} A block diagram illustrating an example of the configuration of an information analysis device according to a second exemplary embodiment of the present invention.
{FIG. 4} An explanatory diagram illustrating an example of documents stored in a document storage section.
{FIG. 5} An explanatory diagram illustrating an example of the result of plotting the results of searching the document storage section in chronological order.
{FIG 6} An explanatory diagram illustrating another example of the result of plotting the results of searching the document storage section in chronological order.
{FIG. 7} A block diagram illustrating an example of the minimum configuration of an information analysis device.
{FIG. 8} A block diagram illustrating the configuration of a search system according to the present invention.
Hereinafter, an exemplary first embodiment of the present invention will be described with reference to the accompanying drawings. The present invention relates to an information analysis device that uses an information analysis method to evaluate the credibility of a linguistic expression as an analyzation object at a specific point in time.
In the present exemplary embodiment, the information analysis device includes a related expression generation section that accepts an evaluation object expression being an linguistic expression to be evaluated, as input and generates a linguistic expression related to the evaluation object expression as a related expression. The information analysis device also includes a credibility calculation section that acquires the evaluation object expression and the related expression from a plurality of documents (electronic documents) along with time information and calculates credibility concerning the meaning of the evaluation object expression at a specific point in time by comparing the number of times that the evaluation object expression appears and the number of times that the related expression appears at the same time point.
With the above configuration employed, the information analysis device calculates credibility by taking into account not only the number of times that the evaluation object expression as an expression to be analyzed appears at each point in time but also the number of times that the related expression, whose appearance correlates with the appearance of the evaluation object expression, appears at each point in time. Therefore, it is possible to appropriately evaluate the credibility of the linguistic expression at a specific point in time.
In the information analysis device illustrated in
The storage device 30 includes a document storage section 31.
The document storage section 31 stores document data, such as articles or blogs on the Internet, along with time information. Suppose that, in an exemplary case, time information, such as a date and time of creation or a date and time of sending the document, is given to the document data on a per-document basis. The time information may be given to a specific portion of the document data. For example, the time information “January 10, 2008” may be given to the linguistic expression portion “XX Corporation announced that . . . on January 10, 2008” inside a given piece of document data. It is not necessary for the document storage section of the storage device to retain documents inside the information analysis device. As long as an access means for accessing the documents is provided, the real documents may be kept outside the information analysis device, not inside the information analysis device. For example, instead of the storage device, a blog search engine may be used to search blogs on the Internet with a keyword and date and time specified.
The data processing device 20 includes an related expression generation section 21 and a credibility calculation section 22.
The related expression generation section 21 is equipped with a function of accepting an evaluation object expression as a linguistic expression whose credibility is to be evaluated as a proposition, from the input device 10 as input. The related expression generation section 21 is also equipped with a function of generating, as an related expression, a linguistic expression related to the evaluation object expression input.
As related expressions, the following expressions can be used: a linguistic expression that is in conflict with the evaluation object expression input and a linguistic expression that is in co-occurrence with the evaluation object expression in the document stored in the document storage section 31.
Equipped with the above functions, the related expression generation section 21, for example, generates an expression whose meaning is in conflict with the evaluation object expression as a related expression. Moreover, the related expression generation section 21, for example, generates an expression in co-occurrence with the evaluation object expression as a related expression by referring to a plurality of electronic documents.
The credibility calculation section 22 is equipped with a function of acquiring (extracting) the search results that correspond to the evaluation object expression input and the related expression generated by the related expression generation section 21 from the document storage section 31 along with time information. The credibility calculation section 22 is also equipped with a function of comparing, based on the time information, the number of times that the evaluation object expression appears with the number of times that the related expression appears at the same time point to calculate the credibility of the evaluation object expression at a specific point in time concerning whether the meaning of the evaluation object expression is true or not.
Equipped with the above function, the credibility calculation section 22 acquires the evaluation object expression and the related expression from a plurality of electronic documents along with time information and compares the number of times that the acquired evaluation object expression appears with the number of times that the acquired related expression appears at the same time point to calculate credibility concerning the meaning of the evaluation object expression at a specific point in time.
A specific point in time at which credibility is calculated is, in an exemplary case, a current or previous point in time. However, if prediction or the like enables the number of times that the evaluation object expression appears and the number of times that the related expression appears to be obtained, a specific point in time may be a future point in time. The credibility calculation section 22, for example, calculates credibility based on the difference between the number of times that the evaluation object expression appears and the number of times that the related expression appears at each point in time or based on the difference between a change in the number of times that the evaluation object expression appears and a change in the number of times that the related expression appears at each point in time.
The credibility calculation section 22 is also equipped with a function of acquiring the evaluation object expression and the related expression from a plurality of electronic documents along with time information and comparing the trend of changes in the number of times that the acquired evaluation object expression appears and the trend of changes in the number of times that the acquired related expression appears at the same time point to make adjustments to the credibility of the evaluation object expression at a specific point in time.
For example, when an upward trend of the evaluation object expression and a downward trend of the related expression are observed at the same time point, the credibility calculation section 22 raises the credibility of the evaluation object expression for a time when the upward trend of the evaluation object expression and the downward trend of the related expression are observed at the same time point or for a time point after the observed time, thereby making adjustments to the credibility. For example, when a downward trend of the evaluation object expression and an upward trend of the related expression are observed at the same time point, the credibility calculation section 22 decreases the credibility of the evaluation object expression for a time when the downward trend of the evaluation object expression and the upward trend of the related expression are observed at the same time point or for a time point after the observed time, thereby making adjustments to the credibility.
The credibility calculation section 22 may be equipped with a function of making adjustments to the credibility of the evaluation object expression at each point in time in accordance with the ratio of the number of times that the evaluation object expression appears to the number of times that the related expression appears at each point in time. The credibility calculation section 22 may be equipped with a function of calculating the rate to which the author of the evaluation object expression is identical with the authors of the related expressions for the evaluation object expression and related expressions acquired from a plurality of electronic documents to make adjustments to the credibility of the evaluation object expression depending on the rate of how many calculated authors are identical. The credibility calculation section 22 may be equipped with a function of calculating the rate to which the source of the evaluation object expression is identical with the sources of the related expressions for the evaluation object expression and related expressions acquired from a plurality of electronic documents to make adjustments to the credibility of the evaluation object expression depending on the rate of how many calculated sources are identical.
The following describes operations;
One of the related expressions generated by the related expression generation section 21 is a linguistic expression in conflict with the input evaluation object expression. In an exemplary case, the conflicting linguistic expression can be generated by adding a negative expression to the input evaluation object expression.
For example, if the evaluation object expression “green tea works well for cancer” is input, the related expression generation section 21 adds the negative expression “not” to the evaluation object expression and then makes adjustments, those common in the field of natural language generation technology or the like, to the conjugation. Therefore, it is possible to generate the conflicting linguistic expression “green tea does not work well for cancer” for the evaluation object expression. Moreover, the related expression generation section 21 may add the negative expression “not” to the linguistic expression “‘A’ is necessary” to generate the conflicting expression “‘A’ is not necessary.” Making use of knowledge about linguistic expressions including the fact that the linguistic expression “necessary” is in conflict with the linguistic expression “unnecessary,” the related expression generation section 21 may generate the conflicting expression “‘A’ is unnecessary.”
The related expression generation section 21 may generate an expression that is likely to be in co-occurrence with the evaluation object expression as an related expression. When an linguistic expression that is likely to be in co-occurrence with the evaluation object expression is generated, the related expression generation section 21, in an exemplary case, searches documents stored in the document storage section 31 on the basis of the input evaluation object expression and extracts a proposition portion that appears near the evaluation object expression from among the obtained results of searching. Then, the related expression generation section 21 can generate the linguistic expression that is likely to be in co-occurrence with the evaluation object expression on the basis of the proposition portion extracted.
For example, suppose that the evaluation object expression “earthquake-proof gel is effective” is input and that the related expression generation section 21 finds a document including the linguistic expression “the use of earthquake-proof gel along with earthquake-resistant pole is effective in withstanding earthquakes” from the document storage section 31 on the basis of the input evaluation object expression. In this case, the related expression generation section 21 extracts the proposition portion “along with earthquake-resistant pole” that appears near the evaluation object expression in the result of searching. Then, the related expression generation section 21 makes adjustments, those common in the field of language generation technology or the like, to the conjugation to generate the linguistic expression “along with earthquake-resistant pole” that is in co-occurrence with the evaluation object expression.
Subsequently, the credibility calculation section 22 searches the document storage section 31 on the basis of the input evaluation object expression and acquires (extracts) the results of searching including the evaluation object expression along with time information (Step A3 in
The credibility calculation section 22 examines whether an unprocessed related expression is acquired (i.e. whether there is an unprocessed related expression) (Step A5 in
Moreover, the credibility calculation section 22 compares the evaluation object expression with related expression in number of times of appearances or in the trend of time change of the number of times of the appearances (Step A7 in
For example, the credibility calculation section 22 can calculate credibility at a given point in time by calculating the difference between the number of times that the evaluation object expression appears and the number of times that the related expression appears at the above point in time and normalizing the resultant value.
The larger the value of the credibility obtained in the above way, the more reliable the contents of the evaluation object expression. When the value of the credibility is below zero, the contents of the evaluation object expression is unreliable. When the value of the credibility is zero, it is not clear whether the contents of the evaluation object expression is reliable, which means that it is not possible to determine only with the documents stored in the document storage section 31.
The credibility calculation section 22 may adjust credibility at a given point in time by further calculating the difference between a change in the number of times that the evaluation object expression appears and a change in the number of times that the related expression appears at the above point in time and adding the value obtained by normalizing the value of the difference or by doing other things. Therefore, even if the credibility calculated from the difference between the number of times that the evaluation object expression appears and the number of times that the related expression appears has the same value, the credibility becomes larger when the evaluation object expression is on an upward trend, while the credibility becomes smaller when the evaluation object expression is on a downward trend. In particular, by comparing the upward/downward trends of the related expressions and measuring the upward/downward trends of the evaluation object expressions, the phenomenon of the evaluation object expressions increasing/decreasing due to the effects of all the related topics that are on the increase/decrease can be expected to be removed.
When an upward trend of the evaluation object expression and a downward trend of the related expression are observed at the same time point, the credibility calculation section 22 may adjust the credibility in a such way that the credibility of the evaluation object expression becomes larger after the above time point. Alternatively, when a downward trend of the evaluation object expression and a upward trend of the related expression are observed at the same time point, the credibility calculation section 22 may adjust the credibility in such a way that the credibility of the evaluation object expression becomes smaller after the above time point.
When a linguistic expression that is in co-occurrence with the input evaluation object expression is generated by the related expression generation section 21 as a related expression, the credibility calculation section 22 may adjust the credibility at a given point in time according to the ratio of the number of times that the evaluation object expression appears to the number of times that the related expression appears at the above point in time. For example, the credibility calculation section 22 makes adjustments in a way that decreases the credibility when the value of the ratio is large and increases the credibility when the value of the ratio is small.
The credibility calculation section 22 may be so configured as to consider that there is a strong correlation between the evaluation object expression and the related expression and increase the adjustable range of the credibility concerning the related expression when the author or source of the evaluation object expression is identical with the author or source of the related expression.
The credibility calculation section 22 then returns to step A5 in
When there is no unprocessed related expression (i.e. all the related expressions have been processed) at step A5 in
As for the way to output credibility, the credibility calculation section 22 may be so configured as to make the output device 4 output credibility at a point in time specified by a user or to make the output device 4 output the time-series changes of credibility for a given period of time in the form of a graph. The credibility calculation section 22 may be so configured as to present (for example, display) related information about what has caused credibility to change dramatically, instead of presenting credibility directly to a user.
As described above, according to the present exemplary embodiment, credibility is calculated by taking into account not only the number of times that the evaluation object expression to be analyzed appears at each point in time but also the number of times that the related expression, whose appearance correlates with the appearance of the evaluation object expression, appears at each point in time. Therefore, it is possible to appropriately evaluate the credibility of a linguistic expression as an analyzation object at a specific point in time.
Moreover, according to the present exemplary embodiment, instead of merely presenting credibility at a specific point in time to a user, the information analysis device may be so configured as to make a determination as to whether the evaluation object expression is reliable at each point in time using a predetermined threshold value and output the period of validity of the evaluation object expression.
Furthermore, according to the present exemplary embodiment, the information analysis device is so configured as to find the related expression that correlates closely with the evaluation object expression at each point in time and make adjustments to the credibility on the basis of the presence of the related expression or time-series changes. Therefore, the configuration enables related information about what has caused the credibility to change dramatically to be presented to a user when the credibility changes dramatically. Thanks to the above configuration, it is possible to help a user to determine credibility by offering appropriate information.
Hereinafter, an exemplary second embodiment of the present invention will be described with reference to the accompanying drawings.
Here, an information analysis program 60 is loaded into the data processing device 50 and controls the operation of the data processing device 50. The information analysis program 60 is also equipped with a function of generating a document storage section 32 in the storage device 30. Moreover, the data processing device 50 is equipped with a function of carrying out the same processes as the data processing device 20 of the first exemplary embodiment does under the control of the information analysis program 60.
Equipped with the above functions, the data processing device 50 of the present exemplary embodiment reads the information analysis program 60 stored in a computer-readable information recording medium, such as a flexible disk like FD (floppy disk), a CD-ROM, a DVD or a flash memory, and carries out processes in accordance with the information analysis program 60 that the data processing device 50 has read. For example, the data processing device 50 carries out processes by reading the information analysis program 60 that enables the following processes to be executed: a related expression generation process of accepting the evaluation object expression, a linguistic expression to be evaluated, as input and generating a linguistic expression related with the evaluation object expression as a related information and a credibility calculation process of calculating credibility concerning the meaning of the evaluation object expression at a specific point in time by acquiring the evaluation object expression and the related expression from a plurality of electronic documents along with time information and comparing the number of times that the acquired evaluation object expression appears and the number of times that the acquired related expression appears at the same time point.
The following describes an example of the present invention with reference to the accompanying drawings. Incidentally, an information analysis device shown in the present example corresponds to the information analysis device of the first exemplary embodiment. In the present example, the information analysis device includes a keyboard as the input device 10 as well as a central processing unit as the data processing device 20. The information analysis device is also equipped with a magnetic disc recording device as the storage section 30 as well as a display, such as a liquid crystal display, as the output device 40. A computer that runs in accordance with the information analysis program constitutes the information analysis device. The information analysis program may be stored in a portion of the storage device 30 or in a memory that the data processing device 20 has. If the information analysis device is a server connected to terminals through a network, the input device 10 and the output device 40 serve as a communication section to communicate with terminals and the keyboard and the display can be omitted.
The central processing unit that the data processing device 20 is equipped with functions as the related expression generation section 21 and the credibility calculation section 22. In the magnetic disc storage device, there is a storage area that functions as the document storage section 31.
For example, in the example illustrated in
The following looks at the case in which the linguistic expression “Pluto is a planet” is input as an evaluation object expression.
The central processing unit accepts, in accordance with an operation performed by a user, the linguistic expression “Pluto is a planet” as an evaluation object expression (which is input through the keyboard). The central processing unit then generates a related expression related to the evaluation object expression on the basis of the input evaluation object expression. For example, assume that a linguistic expression in conflict with the evaluation object expression will be generated as a related expression. In this case, the central processing unit adds the negative expression “not” to the linguistic expression “Pluto is a planet” and makes adjustments to the conjugation to generate the related expression “Pluto is not a planet.”
The central processing unit then searches the document storage section 31 on the basis of the evaluation object expression “Pluto is a planet” (with the key “Pluto is a planet”). Subsequently, the central processing unit takes out the related expression “Pluto is not a planet” as an unprocessed related expression and searches the document storage section 31 on the basis of the related expression (with the key “Pluto is not a planet”).
For a search method, the central processing unit may adopt a method capable of acquiring a wide range of expressions with the same meaning and the like, which are those expressing the same contents in terms of meaning, to search the document storage section 31, instead of extracting only an expression that is exactly identical with the evaluation object expression or related expression simply as a character string. For example, the central processing unit may adopt a method capable of acquiring a wide range of synonymous expressions (expressions with the same meaning) and the like, which are those expressing the same contents in terms of meaning, including expressions that are the same in sentence structure as the evaluation object expression or related expression and expressions created by replacing part of the evaluation object expression or related expression with a synonym.
Moreover, for the search results of the evaluation object expression and the related expression, the central processing unit compares the evaluation object expression with the related expression in the number of times of appearances or the trends of the time changes of the number of times of appearances and calculates credibility. For example, the central processing unit can calculate the credibility of the evaluation object expression using the formula “f(t)=α(t)×((the number of times that the evaluation object expression appears)−(the number of times that the related expression appears)). Here, α(t) is a function for outputting a coefficient for normalization at each point “t” in time.
Moreover, the central processing unit can adjust the credibility of the evaluation object expression using the formula “g(t)=β(t)×((the change in the number of times that the evaluation object expression appears)−(the change in the number of times that the related expression appears)). Here, β(t) is a function for outputting a coefficient for normalization at each point “t” in time. In this case, the central processing unit can calculate the adjusted credibility by calculating the weighted sum of f(t) and g(t) or the like.
Furthermore, if expressions whose authors or sources are the same are found after the search results of the evaluation object expression and the related expression are compared, the central processing unit may adjust the credibility according to the ratio in the search results. For example, what the central processing unit applies to each of the above formulas may be the number of times that the related expression appears multiplied by a value greater than or equal to 1 as the number of times that the related expression appearances, depending on the rate of how many authors or sources are identical in the search results. In this case, the credibility is calculated on the assumption that the related expression has appeared a lot more times.
The following looks the case in which the linguistic expression “earthquake-proof gel is effective” is input as an evaluation object expression.
The central processing unit accepts, in accordance with an operation performed by a user, the linguistic expression “earthquake-proof gel is effective” as an evaluation object expression (which is input through the keyboard). The central processing unit then generates a related expression related with the evaluation object expression on the basis of the input evaluation object expression. For example, suppose that an expression that is likely to be in co-occurrence with the evaluation object expression will be generated as a related expression.
The central processing unit then searches the document storage section 31 on the basis of the evaluation object expression “earthquake-proof gel is effective” (with the key “earthquake-proof gel is effective”). If a document including the linguistic expression “the use of earthquake-proof gel along with earthquake-resistant pole is effective in withstanding earthquakes” is found, the central processing device extracts the proposition portion “along with earthquake-resistant pole” that appears near the evaluation object expression in the search results. The central processing unit then makes adjustments of the conjugation, which is generally carried out in the field of language generation technology or the like, to the extracted proposition portion and generates the linguistic expression “along with earthquake-resistant pole” that is in co-occurrence with the evaluation object expression.
Then, in the same way as described above, the central processing unit searches the document storage section 31 on the basis of the evaluation object expression “earthquake-proof gel is effective”. Subsequently, by performing the same process as in the example illustrated in
Moreover, in the same way as in the example illustrated in
Incidentally, when an expression that is likely to be in conjunction with the evaluation object expression is used as a related expression, the central processing unit may calculate, in particular, the ratio of the number of times that the evaluation object expression appears to the number of times that the related expression appears and make adjustments to the credibility according to the value of the ratio. For example, when the value of the ratio of the number of times that the evaluation object expression appears to the number of times that the related expression appears is greater than or equal to a predetermined threshold value, the central processing unit may calculate credibility after replacing the number of times that the related expression appears with a larger value. When the value of the ratio is less than or equal to a predetermined threshold value, the central processing unit may calculate credibility after replacing the number of times that the related expression appears with a smaller value.
As described above, according to the present exemplary embodiment, the central processing unit takes into account not only the number of times that the evaluation object expression, to be analyzed appears at each point in time but also the number of times the related expression, whose appearance correlates with the appearance of the evaluation object expression, appears at each point in time when calculating credibility. Therefore, it is possible to appropriately evaluate the credibility of a linguistic expression as an analyzation object at a specific point in time.
The following describes the minimum configuration of an information analysis device according to the present invention.
In the information analysis device having the minimum configuration as illustrated in
The information analysis device having the minimum configuration as illustrated in
The information analysis device of each of the above exemplary embodiments can be applied for the use of a search system that presents, as a search condition, an expression that is input through the information analysis device and considered to be highly reliable.
The document search section 70 receives, as a search condition, a highly reliable expression output from the information analysis device 100 and searches a plurality of documents that can be accessed by the search object document database 90 for a document containing the received expression. The document output device 80 outputs a document found by the document search section 70. The search object document database 90 is a database that enables a search object collection of documents to be accessed. The search object document database 90 may be formed as a database that offers access to a collection of documents such as Internet texts. A collection of documents to be searched may be stored in the search object document database 90. Alternatively, only an access means, such as URL, for accessing each document may be provided, with documents themselves being stored outside.
The above has described the exemplary embodiments of the present invention. However, the present invention may be embodied in other forms without departing from the spirit and essential characteristics defined by the appended claims. The described embodiments are therefore to be considered only as illustrative, not as restrictive. The scope of the invention is indicated by the appended claims, not by the specification or the abstract. Furthermore, all modifications and alterations which come within the meaning and range of equivalency of the claims are to be embraced within the scope of the present invention.
The present application claims priority from Japanese Patent Application No. 2008-019015 filed on Jan. 30, 2008, the entire contents of which being incorporated herein by reference.
The present invention can be applied for the use of an information analysis device that evaluates the credibility of an input proposition using information on the Internet. Moreover, the present invention can be applied for the use of a program that realizes the information analysis device with a computer.
10: Input device
20, 50: Data processing device
30: Storage device
40: Output device
60: Information analysis program
21: Related expression generation section
22: Credibility calculation section
31, 32: Document storage section
| Number | Date | Country | Kind |
|---|---|---|---|
| 2008-019015 | Jan 2008 | JP | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/JP2009/051538 | 1/30/2009 | WO | 00 | 7/28/2010 |
