JP2002175330A - Information retrieval device, score-determining device, method for retrieving information, method for determining score, and program recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically retrieve estimation information related to a retrieving word (e.g., a commodity name) specified by a user from a web page or the like over the Internet, and to successively present retrieved information starting from the one most which appears to be reputation information, to the user. SOLUTION: A document input means 11 inputs a document from a web page or the like over the Internet and a retrieving word input means 12 inputs a retrieval word instructed by a user. A reputation information discrimination means 13 extracts reputation information included in the inputted document and concerned with the instructed retrieving word, on the basis of the appearing position of the retrieving word in the document and the appearing position of evaluation expression for determining the value of the matter and a reputation information priority judging means 16 places priorities on respective pieces of reputation information in the order of similarity to reputation information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information search technique, and more particularly to an information search technique capable of searching for reputation information.

[0002]

2. Description of the Related Art At present, there are many information retrieval systems which automatically collect attribute information based on a search word and present the attribute information to a user. For example, MySimon (http: //www.mysimon.c
The system provided by om) presents attribute information (price) related to a search term (product name) to a user as follows. When the user enters a product name as a search term, a certain online shop α that sells the product
Send product name to In response to this, the online shop α displays the price, size, color,
Search for attribute information such as weight and return the search result in a fixed format. When the attribute information is returned, the price information is extracted based on the format of the online shop α that is recognized in advance, and is presented to the user.

[0003] As a system that can retrieve attribute information based on a search word, there is a system that allows a user to see desired information from attribute information collected manually in advance. For example, Deja.com (http: //www.dej
In a.com), reputation information is collected in advance from consumers as attribute information on products, and the user can obtain the reputation information by selecting the name of the product.

[0004]

However, in the information retrieval system capable of automatically collecting the above-mentioned attribute information, the appearance position can be specified by using standard retrieval result information such as a product name and a price attribute. Even though attribute information can be handled, attribute information such as reputation information, which does not appear as search result information and whose appearance position cannot be specified, cannot be automatically collected.

On the other hand, in an information retrieval system that collects attribute information manually in advance, it is possible to handle unspecified attributes such as reputation information. There is a limit, and even in the reputation information of the above-mentioned products, it is the current situation that products other than extremely prominent products are not enough to be used as judgment materials for purchasing products.

SUMMARY OF THE INVENTION The present invention has been made in view of such a problem, and an object of the present invention is to make it possible to automatically search for reputation information, which is unspecified attribute information, and to use the automatically searched reputation information most. The object is to enable the user to be presented with reputation information in order.

[0007]

In order to achieve the above object, an information retrieval apparatus according to the present invention converts reputation information relating to a specified search term present in an input document into the search term in the document. Reputation information discriminating means (13 in FIG. 1), which is extracted based on the appearance position of the expression and the appearance position of the evaluation expression that determines the value for things, and each reputation extracted by the reputation information discriminating means (13 in FIG. 1). It has reputation information priority determining means (16 in FIG. 1) for assigning priorities to information in the order of reputation information.

[0008] By adopting such a configuration and performing automatic judgment of reputation information priority in addition to automatic extraction of reputation information, a user can preferentially know what is more reputation information. .

In the information retrieval apparatus of the present invention, the reputation information discriminating means (13a in FIG. 12) specifies a field in which the search word is used in order to improve the accuracy and speed of extracting reputation information. Search term field specifying means (13 in FIG. 12)
6) and search word position specifying means (13 in FIG. 12) for specifying a search word appearance position where the search word appears in the document.
2) and an evaluation expression storage unit (1 in FIG. 12) storing a plurality of evaluation expressions and field information indicating a field of use of each evaluation expression.
31a) and in the field specified by the search term field specifying means (136 in FIG. 12) of the evaluation expressions stored in the evaluation expression storage unit (131a in FIG. 12) in the document. Evaluation expression position specifying means for specifying the evaluation expression appearance position where the evaluation expression appears (133a in FIG. 12)
The search word appearance position specified by the search word position specifying means (132 in FIG. 12) and the evaluation expression position specifying means (FIG. 1)
An inter-expression distance determining means (134 in FIG. 12) for obtaining a pair of a search word and an evaluation expression within a certain distance from each other based on the evaluation expression appearance position specified by 2133a);
Reputation information extraction means (135 in FIG. 12) for extracting reputation information including at least a portion sandwiched between the search term and the evaluation expression constituting the pair obtained by the expression distance determination means (134 in FIG. 12). ).

[0010] By adopting such a configuration, when the field in which the search word is used is, for example, "book", an evaluation expression such as "delicious" which is not used for reputation information on "book" is excluded. Since reputation information extraction processing can be performed, the accuracy of reputation information extraction and the extraction processing speed are improved.

[0011] In addition, the score determination device of the present invention, when assigning priorities to reputation information, in order to more accurately assign priorities, designates a specified search term existing in an input document. Reputation information determining means (143 in FIG. 14) for extracting reputation information on the basis of the appearance position of the search word in the document and the appearance position of an evaluation expression that determines value for things; Each piece of reputation information extracted by the reputation information discriminating means for determination (143 in FIG. 14) is displayed on the display device and presented to the user, and based on the score for each piece of reputation information input by the user using the input device. Score determining means (146 in FIG. 14) for determining the score to be stored in the score storage unit (163 in FIG. 4).
And

By adopting such a configuration, the score input by the user is used for the extracted reputation information,
Since the score to be stored in the score storage unit (score used when determining the priority order) is determined,
When assigning priorities to reputation information, it is possible to assign priorities more accurately.

[0013]

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

[First Embodiment] Referring to FIG.
The first embodiment of the information search device 1 according to the present invention includes a document input unit 11, a search word input unit 12, a reputation information determination unit 13, a determination result storage unit 14, a reputation information storage unit 15, And a reputation information priority determination unit 16, a priority storage unit 17, an input device 2 such as a keyboard, a display device 3 such as an LCD, and a recording medium K1. The information search device 1 is configured by a computer.

The document input means 11 has a function of inputting a document set which is a set of documents to be searched for reputation information and passing it to the reputation information discriminating means 13. The input destination of the document may be a Web page on the Internet or a file in which a plurality of documents are recorded.

The search word input means 12 has a function of receiving a search word input by the user using the input device 2 and passing it to the reputation information determination means 13.

The reputation information discriminating means 13 includes the document input means 1
1 has a function of receiving a document set from the search term input unit 1, receiving a search term from the search term input unit 12, and extracting reputation information related to the search term included in the document set.

As shown in FIG. 2, reputation information discriminating means 13
Is the evaluation expression storage unit 131 and the search word position specifying unit 13
2, an evaluation expression position specifying unit 133, an inter-expression distance determining unit 134, and a reputation information extracting unit 135.

The evaluation expression storage section 131 stores a plurality of evaluation expressions. Here, the evaluation expression is an expression that can determine the value for an object. Therefore, expressions that do not show value for things such as "play", "move", and "research" are not evaluation expressions. Evaluation expression is "good"
In addition to words like "like", "best" and "funny",
There are also nouns such as "popular", "exquisite" and "wasteful". As for the storage format of the evaluation expressions in the evaluation expression storage unit 131, a method of storing in a unit that absorbs a difference in notation by utilizing each evaluation expression is effective in terms of saving storage capacity.
It is not limited to this method.

The search word position specifying means 132 receives the document set from the document input means 11, receives the search word from the search word input means 12, and specifies the appearance position of the search word (the search word appearance position) in the document set. Also, it has a function of outputting the specified search word appearance position to the inter-expression distance determination unit 134. The search word appearance position can be specified by using a method of calculating a distance from a reference position such as the beginning or end of a document. Further, the method of calculating the distance from the reference position may be any method as long as the distance from the reference position can be calculated. Specifically, there are a method of calculating the distance based on the number of characters from the head of the document, a method of calculating the distance based on the number of bytes from the head of the document, and a method of calculating the distance based on the number of lines from the head of the document. Also,
The position information of the search term to be output includes at least information for specifying a document in which the search term appears, and information for specifying a position in the document in which the search term appears.

The evaluation expression position specifying means 133 receives the document set from the document input means 11 and
1 and a function for specifying a position (evaluation expression appearance position) where any of the evaluation expressions stored in the evaluation expression storage unit 131 in the document set appears, and a function for specifying the specified evaluation expression appearance position as a distance between expressions. It has a function of outputting to the determination means 134. Specifically, evaluation expression position specifying means 1
33 checks, for each evaluation expression stored in the evaluation expression storage unit 131, whether the evaluation expression is included in the document set, and if the evaluation expression is included in the document set, determines the appearance position of the evaluation expression. Identify. The specification of the appearance position of the evaluation expression in the evaluation expression position specifying means 133 is performed by the search word position specifying means 1
32 in the same manner. Further, the output position of the evaluation expression to be output includes at least information for specifying the document in which the evaluation expression appears, and information for specifying the position of the evaluation expression in the document.

The expression-to-expression distance determining means 134 determines whether the distance between the expressions is within a predetermined range based on the search word appearance position specified by the search word position specifying means 132 and the evaluation expression appearance position specified by the evaluation expression position specifying means 133. Function to find a pair of search words and evaluation expressions (search word / evaluation expression pairs)
It has a function of storing the search words, evaluation expressions, and their appearance positions that make up the evaluation expression pair in the determination result storage unit 14. Here, the distance is within a certain range, for example, when the magnitude of the numerical difference between the search word and the evaluation expression is equal to or less than the threshold,
This applies when there is a parsing dependency between the search word and the evaluation expression. The appearance position stored in the determination result storage unit 14 includes information for specifying the document and information for specifying the appearance position in the document. In the present embodiment, since a pair of a search word and an evaluation expression within a certain distance from each other is set as a search word / evaluation expression pair, a certain distance between the search word and the evaluation expression is fixed. When there are a plurality of evaluation expressions within the range, each pair of the certain search word and the plurality of evaluation expressions is a search word / evaluation expression pair. May be used as a search word / evaluation expression pair only. At this time, it is necessary to select one evaluation expression from among a plurality of evaluation expressions. For example, a method of selecting the one farthest from the search word, a method of selecting the one closest to the search word, A method of selecting the last one can be adopted.

The reputation information extraction means 135 determines at least the search word / word based on the appearance position of the search word / evaluation expression pair stored in the discrimination result storage unit 14 and the document set input from the document input means 11. It has a function of extracting reputation information including a character string of a portion sandwiched between evaluation expression pairs and storing the extracted reputation information in the reputation information storage unit 15. The reputation information to be extracted may be, for example, only a character string sandwiched by the search word / evaluation expression pair, or may exist before and after the character string sandwiched by the search word / evaluation expression pair. It may be a combination of a predetermined number of characters and a character string.

The reputation information priority determining means 16 assigns priorities to reputation information stored in the reputation information storage unit 15 in the order of reputation information, and stores the reputation information in the reputation information storage unit 15. It has a function of displaying each piece of reputation information in the order of priority on the display device 3 and a function of storing the order of priority for each piece of reputation information in the priority order storage unit 17.

Referring to FIG. 4, reputation information priority determining means 16 includes a rule storage section 161 and a rule matching section 16.
2, a score storage unit 163, and a score search unit 164
, An alignment unit 165, and a presentation unit 166.

The rule storage unit 161 stores rules that are criteria for determining the priority of reputation information. FIG. 6 shows an example of the contents of the rule storage unit 161. In the example of FIG. 6, a plurality of rules regarding a positional relationship between a search word and an evaluation expression and a plurality of rules regarding a position of the evaluation expression in a sentence are stored. Rules relating to the positional relationship between the search word and the evaluation expression include rule ID1 “the search word and the evaluation expression are in different sentences” and rule ID2 “the search word and the evaluation expression are close to each other”. As rules relating to the position of the evaluation expression in the text, the evaluation expression appears in the vicinity of the expression indicating guesswork (eg, guesswork, etc.) in the rule ID11, and the evaluation rule appears in the vicinity of the front of the question mark in the rule ID12. An expression appears ". Although not shown in the example of FIG. 6, a rule regarding the position of the search word in the text may be used. As specific examples of this rule, "a search word appears before an expression of a contradiction (such as a burn)," or "a search word appears before an expression of a cause (such as, a so)" For the rule description method, regular expressions that can express the appearance pattern of character strings are effective, but it is also possible to specify the part of speech and original form of words by morphological analysis and the dependency between words by syntax analysis as conditions It is.

The rule matching means 162 stores the reputation information stored in the reputation information storage unit 15 in the rule storage unit 1.
61, a value indicating whether the reputation information satisfies the rule for each of the rules (in the present embodiment, “1” if satisfied, unsatisfied In this case, “0” is output. Here, when the rule specifies the part of speech and original form of the word by morphological analysis and the dependency between words by syntactic analysis as conditions, morphological analysis and syntactic analysis are used for the character string.

The score storage unit 163 determines the priority of reputation information for each combination of values ("1" or "0") for each rule, which may be output from the rule matching unit 162. Stores the score to be used. More specifically, values for each rule are stored in an array format, and scores are stored in association with each array. Here, as the method of storing the score,
Register as Key and score as Value. An example is shown in FIG. Note that the example of FIG. 7 is for the case where 12 rules as shown in FIG. 6 are adopted as the rules for determining the priority of the reputation information. The score is
The highest is 5 and the lowest is 1, and the higher the value, the higher the priority.

The score retrieving means 164 converts an array having the values for each rule (12 values for 12 rules) output from the rule matching means 162 as key elements into a key.
By searching the score storage unit 163, a function of searching for a score for the reputation information is provided.

The arranging means 165 includes a score searching means 164
A function of assigning priorities to information (reputation IDs) for specifying the respective pieces of reputation information in the descending order of the score, based on the scores for the respective pieces of reputation information stored in the reputation information storage unit 15 retrieved by Reputation ID after prioritization
To the presentation means 166 and a function of storing the reputation ID after the priority is assigned in the priority storage 17.

The presentation means 166 sorts each piece of reputation information stored in the reputation information storage unit 15 in the descending order of priority (reputation information) in accordance with the reputation ID with the priority passed from the sorting means 165. A function of displaying on the display device 3 in descending order of the probability).

The recording medium K1 included in the information search device 1
Is a disk, a semiconductor memory, or another recording medium in which a program for causing a computer to function as the information search device 1 is recorded. This program is
By being read by a computer and controlling its operation, a document input unit 11, a search word input unit 12, a reputation information determination unit 13, and a reputation information priority determination unit 16 are realized on the computer.

[Description of Operation] Next, the operation of the first embodiment of the present invention will be described with reference to the drawings. The operation will be described in two parts: a part for extracting reputation information and a part for prioritizing the extracted reputation information.

[1. Operation at the time of reputation information extraction] First, the document input means 11 inputs a document set to the reputation information discriminating means 13, and the search word input means 12 inputs a character string serving as a search word of reputation information (S31 in FIG. 3). ). When inputting a document set to the reputation information determining means 13, the document input means 11 assigns a unique document number to each document included in the document set, and identifies each document in the document set by the document number. I can do it.

Next, the search word position specifying means 132 in the reputation information discriminating means 13 specifies the appearance position of the search word in the document set, and searches the search word appearance position information including the specified appearance position and the search word between the expressions. The data is output to the distance determining means 134 (S32).

Now, for example, in the document set input from the document input means 11, as the document of document number 1, "Product A is very interesting. Product B is recommended because its price is low."
, And "commodity A" is input as a search word from the search word input means 12, the search word position specifying means 1
32 performs the following processing on the document with document number 1 (the search term position specifying means 132
By from the beginning of the document as a method for specifying the search word appearance
It is assumed that a method of calculating the number of te is adopted). First,
The appearance position of the search term “product A” is specified as the 0th byte from the head of the document of document number 1. Next, search word appearance position information “document number 1, product A, 0 Byte” including search word appearance position “document number 1, 0 Byte” and search word “product A”
e ”is generated and output to the expression distance determination means 134.

When the process of S32 is completed, the evaluation expression position specifying means 133 specifies the appearance position of each evaluation expression stored in the evaluation expression storage unit 131 in the document set, and specifies the specified appearance position and evaluation expression. Is output to the expression-to-expression distance determining means 134 (S3).
3). Now, for example, “funny, recommended, favorite, good” is stored in the evaluation expression storage unit 131 as a set of evaluation expressions, and the evaluation expression position specifying unit 133 uses the evaluation expression Assuming that the method of calculating the number of bytes from the head is adopted, the evaluation expression position specifying means 133 performs the following processing on the document of document number 1. First, the evaluation expression position specifying unit 133 recognizes that “interesting” and “recommended” are included in the document of the document number 1 among the evaluation expressions stored in the evaluation expression storage unit 131, and The appearance positions of "document number 1, 14 Byte" and "document number 1, 48 By"
te ". Then, the evaluation expression appearance position information “document number 1, interesting, 14 Byte” including the evaluation expression appearance position “document number 1, 14 Byte” and the evaluation expression “funny”
e ”and the evaluation expression appearance position“ document number 1, 48 Bytes ”
e ”and evaluation expression appearance position information“ document number 1, recommended, 48 Bytes ”including the evaluation expression“ recommended ”are output to the inter-expression distance determination unit 134.

The processing order of S32 and S33 can be reversed.

Next, the expression-to-expression distance determining means 134 calculates the search term appearance position information output from the search word position specifying means 132 and the evaluation expression appearance position information output from the evaluation expression position specifying means 133, A pair (search word / evaluation expression pair) of the search word and the evaluation expression within a certain distance from each other is obtained, and the pair appearance position information including the search word, the evaluation expression, and their appearance positions constituting the pair is obtained. It is stored in the determination result storage unit 14 (S34, S35).

Now, for example, the search word appearance position "document number 1, product A, 0 Byte" is sent from the search word position specifying means 132.
Is output, and the evaluation expression appearance position information “document number 1, interesting, 14 bytes” is output from the evaluation expression position specifying unit 133.
Assuming that "document number 1, recommended, 48 bytes" is output, the following processing is performed in S34 and S35. First, the distance between the search word “product A” and the evaluation expressions “funny” “recommended” is obtained. In this example, the distance between the search word “product A” and the evaluation expression “funny” is 14−
0 = 14 bytes, and the distance between the search term “product A” and the evaluation expression “recommended” is 48−0 = 48 bytes. Thereafter, it is checked whether or not the distance between the search word and the evaluation expression is equal to or smaller than a predetermined threshold (30 bytes in this example).
The paired appearance position information including the evaluation expression and its appearance position is stored in the determination result storage unit 14. In this example, the distance between the search word “product A” and the evaluation expression “funny” is equal to or less than the threshold, and the distance between the search word “product A” and the evaluation expression “recommended” exceeds the threshold. Therefore, the expression distance determination means 13
Reference numeral 4 denotes pair appearance position information (document number 1, product A, 0 Byte, interesting,
14 Bytes).

In the above-described operation, the description has been given of the method in which the expression-to-expression distance determining means 134 starts the operation after information is output from both the search position specifying means 132 and the evaluation expression position specifying means 133. In addition, it is also possible to perform the operation based on one of the position of the search word specified by the search word position specifying unit 132 and the position of the evaluation expression specified by the evaluation expression position specifying unit 133. Now, a case where the position information of the search word specified by the search word position specifying means 132 is used as a reference will be described. First, the search word position specifying means 132, which is the reference means, uses the search word appearance position information as the distance between expressions. The search term output from the search term position specifying means 132, which is the reference means, is output to the evaluation expression position specifying means 133, which is not used as the reference. Outputs appearance position information. Then, the evaluation expression position specifying means 133, which is not used as a reference, uses the received search word appearance position and the range of distance (threshold) to move forward or backward by the range of the distance set from the position of the search word. It checks whether or not there is an evaluation expression, and if there is an evaluation expression, specifies the position of the evaluation expression and outputs evaluation expression appearance position information of the evaluation expression to the inter-expression distance determination means 134. Then, finally, the expression-to-expression distance determination means 134 receives the evaluation expression appearance position information output from the evaluation expression position identification means 133 which is a means not used as a reference, and, based on the search word appearance position information and the evaluation expression appearance position information, When the distance between the search word and the evaluation expression is within a certain range, a method of storing the pair appearance position information in the determination result storage unit 14 may be used.

Thereafter, the reputation information extracting means 135 uses at least the search word / evaluation expression based on the pair appearance position information stored in the discrimination result storage unit 14 and the document set input from the document input means 11. The character string sandwiched between the pairs is extracted as reputation information and stored in the reputation information storage unit 15 (S36, S37). Now, for example, assuming that the storage content of the determination result storage unit 14 is as shown in FIG. 8, reputation information as shown in FIG. 9 is stored in the reputation information storage unit 15.

[2. Operation at the time of giving priority] Now, the rules serving as criteria for determining the priority of the reputation information stored in the rule storage 161 are the twelve rules shown in FIG. 6 and the contents of the score storage 163 are shown in FIG. The operation will be described assuming that shown in FIG. 8, FIG. 9, FIG. 10, FIG.
The correspondence of 1 is linked by reputation ID.

The rule matching means 162 in the reputation information priority determination means 16 performs the following processing for each piece of reputation information stored in the reputation information storage unit 15. One piece of reputation information currently being processed and a rule storage unit 161
To each of the 12 rules stored in
For each of the rules, a value indicating whether the reputation information to be processed satisfies the rule and the reputation ID of the reputation information to be processed are output (S51 in FIG. 5).

Next, the score search unit 164 searches the score storage unit 163 using the array having the twelve values output from the rule matching unit 162 as components as a key, thereby searching the score for the reputation information. Then, the searched score and the reputation ID are output (S52). FIG. 10 shows an array for each reputation ID and a result of searching for a score from the score storage unit 163 in FIG. 7 using the key as a key.

Next, the sorting means 165 sorts the plurality of reputation IDs output from the score searching means 164 in the order of the size of the score added thereto, and outputs the order as the priority of the reputation information. (S53). In the present embodiment, priorities are assigned in descending order of the scores, and each reputation ID is assigned.
Are assigned priorities as shown in FIG.

Thereafter, the presenting means 166 is
According to the prioritized reputation ID passed from 5
The reputation information stored in the reputation information storage unit 15 is displayed on the display device 3 in the order of higher priority (in the order of higher probability of being reputation information) (S54).

As described above, in the first embodiment of the present invention, reputation information is automatically extracted, and the priority of reputation information is automatically determined. The user can obtain it preferentially.

[Second Embodiment] Next, a second embodiment of the present invention will be described.
An embodiment will be described. The second embodiment of the present invention differs from the information retrieval apparatus shown in FIG. 1 in that a reputation information discriminating means 13a having a configuration shown in FIG. Is different from the embodiment of FIG.

The reputation information discriminating means 13a shown in FIG.
2 is different from the reputation information discriminating means 13 shown in FIG. 2 in that an evaluation expression storing unit 131a is provided instead of the evaluation expression storing unit 131,
3 in that an evaluation expression position specifying means 133a is provided in place of No. 3, and a search word field specifying means 136 is added.

In the evaluation expression storage unit 131a, in addition to the evaluation expression, field information indicating a field where the evaluation expression is used or a field where the evaluation expression is not used is stored in pairs with the evaluation expression. .

The search word field specifying means 136 has a function of specifying the field of the search word input from the search word input means 12 and outputting the result to the evaluation expression position specifying means 133a. Here, the method of specifying the field of the search word is to specify the input field as the field of the search word when the input of the search word input means 12 includes the search word and the field of the search word. In other cases, the field of the search term is specified by referring to a database (not shown) that stores the correspondence between the search term and the field of the search term.

The evaluation expression position specifying means 133a
A function of selecting only an evaluation expression used in the field specified by the search word field specifying unit 136 from a plurality of evaluation expressions stored in the evaluation expression storage unit 131a, and a document input unit of the selected evaluation expression 11 has a function of specifying an appearance position in the document input from the server 11 and outputting evaluation expression appearance position information including the evaluation expression that has appeared and its appearance position.

[Description of Operation] The operation of the second embodiment will be described with reference to the drawings. The reputation information discriminating means 13
Operations other than a are the same as in the first embodiment,
Here, only the operation of the reputation information determining means 13a will be described.

Referring to FIG. 13, in S31, a document set and a search word are input from the document input means 11 and the search word input means 12, and in the next S32, the search word position specifying means 132 determines the appearance position of the search word. Identify and output search word appearance position information. The operations in S31 and S32 are the same as the operations in S31 and S32 in FIG.

Thereafter, the search word field specifying means 136 specifies the field of the search word (S131). Here, if the field of the search word is also input from the search word input means 12, the input field is specified as the field of the search word. Otherwise, the search field and the field of the search word are specified. The field of the search term is specified by referring to a database (not shown) storing the correspondence.

The evaluation expression position specifying means 133a uses the search word field output from the search word field specifying means 136 to select an evaluation expression used in the search word field from the evaluation expression storage section 131a ( S132). And
The evaluation expression position specifying unit 133a specifies a position where the selected evaluation expression appears in each document, and evaluates expression expression appearance position information on the evaluation expression.
4 (S133). After that, the expression distance determining means 134 and the reputation information extracting means 135 perform the same operation as the operation described in the first embodiment (S34 to S3).
7).

It should be noted that, in the operation of the present embodiment, S31 → S131 → S32 → S1
32 → S133 → S34 → S35 → S36 → S37, S
31 → S131 → S132 → S32 → S133 → S34
→ S35 → S36 → S37, S31 → S131 → S13
2 → S133 → S32 → S34 → S35 → S36 → S3
Similar processing is possible in the order of 7.

In this embodiment, for example, when the field of the search word is “book”, when the evaluation expression position specifying unit 133a specifies the position of the evaluation expression, the evaluation expression storage unit 131a
Out of the evaluation expressions stored in the book, the evaluation expression that is not used in the reputation of the “book” such as the evaluation expression “delicious” can be excluded, and the position of the evaluation expression can be specified. For this reason, it is possible to improve the accuracy of reputation information determination and increase the speed of determination.

[Method of Determining Score] Next, a method of determining a score to be stored in the score storage unit 163 shown in FIG.

Referring to FIG. 14, the score determination device comprises:
Document input means 141 for score determination, search word input means 142 for score determination, and reputation information discriminating means 1 for score determination
43, a score determination determination result storage unit 144, a score determination reputation information storage unit 145, and a score determination unit 146.
And an input device 147 such as a keyboard, and a display device 148 such as an LCD.

The score determining document input means 141 and the score determining search word input means 142 have the same configuration and functions as the document input means 11 and the search word input means 12.

The reputation information determining means 143 for score determination
Reputation information determining means 13 described in the first embodiment, or reputation information determining means 13 described in the second embodiment
It has the same configuration and function as a.

The score determining means 146 displays the reputation information stored in the reputation information storage unit for score determination 145 on the display device 147 and presents it to the user. It has a function of determining the score to be stored in the score storage unit 163 shown in FIG. 7 based on the score for the reputation information input using 148.

Referring to FIG. 15, score determining means 146
Is a score determination rule storage unit 201, a score determination rule matching unit 202, and a frequency score storage unit 203
Score with frequency search means 204, score calculation means 205, display control means 206, score input means 20
7 is provided.

The same content as the rule storage unit 161 described in the first embodiment is stored in the score determination rule storage unit 201.

The rule determining means 202 for score determination has the same structure and function as the rule matching means 162 described in the first embodiment.

The display control means 206 has a function of displaying reputation information stored in the reputation information storage unit for score determination 145 on the display device 147.

The score input means 207 is a display device 147
Has a function of accepting a score (a score indicating the likelihood of reputation information) input by a user who has viewed the reputation information displayed in the input device 148.

In the score storage unit 203 with frequency, in addition to the information stored in the score storage unit 163 of the first embodiment, the sum of the scores from the user input by the score input unit 207 and the score Stores the total number of inputs.
The score stored here is the average value of the scores obtained by dividing the total sum of the scores by the total number of times of inputting the scores.

The score-with-frequency search means 204 outputs a value indicating whether or not the reputation information output from the score determination rule matching means 202 satisfies each rule stored in the score determination rule storage section 201. It has a function of searching the score storage unit 203 with frequency using an array as a component as a key, and searching for the total sum of scores and the total number of times of input of scores for the reputation information.

The score calculation means 205 adds the user's score for the reputation information input from the score input means 207 to the total sum of the scores searched by the frequency score search means 204, and adds After adding 1, the total sum of the updated scores and the total number of input scores are stored in the corresponding portion of the scored frequency storage unit 203 (the portion searched by the scored frequency search means 204), and the total sum of the scores is stored. It has a function to store the average value obtained by dividing the score by the total number of input times.

[Description of Operation] Next, the operation of the score determination device shown in FIG. 14 will be described in detail.

First, the reputation information discriminating means for score determination 14
3 stores the pair appearance position information of the search word / evaluation expression pair in the determination result storage unit for score determination 144, and stores the reputation information in the reputation information storage unit for score determination 145. By this processing, the score determination determination result storage unit 144 is stored in FIG.
7 is stored, and the reputation information shown in FIG. 18 is stored in the reputation information storage unit 145 for score determination. 17 and 18 are indicated by reputation IDs.

When the reputation information is stored in the score determination reputation information storage section 145, the display control means 206 stores the reputation information stored in the score determination reputation information storage section 145 as shown in FIG. The information is displayed on the display device 147, and a score input field is displayed (S161 in FIG. 16).

The user who sees this display inputs a score of reputation information likeness for each piece of reputation information using the score input box (S162). Now, for example, it is assumed that the user inputs the score shown in FIG. 20 for each reputation information.
The correspondence between FIG. 20 and FIG. 17 and FIG.
Indicated by

Next, score determining rule matching means 202
Selects one of the reputation information for which the score has been input from the score input means 207 (S163), and determines the priority order of the reputation information stored in the score determination rule storage unit 201 with respect to this reputation information. Each of the reference rules is collated, and a value indicating whether or not each rule is satisfied is output (S1).
64). It is assumed that the same rule as the twelve rules shown in FIG. 6 is stored in the score determination rule storage unit 201.

Next, the score with frequency search means 204
With the array having the values for the respective rules output from the score determining rule matching means 202 as constituent elements as a key, the total score and the total number of input scores are searched from the score storage unit 203 with frequency (S165).

Next, the score calculation means 205 adds the user's score for the reputation information input from the score input means 207 to the total sum of the scores output from the score search means with frequency 204, and inputs the total score. One is added to the number of times, and then an average value is calculated by dividing the total score by the total number of times of inputting the score (S166). Further, they are stored in the corresponding portion (S
At 165, the total score and the total number of input times are stored in the searched score search unit 204 with frequency (S1).
67).

Finally, if there is an unprocessed input, S
The process returns to S163 (S168). By the above processing, the contents of the score storage unit with frequency 203 are as shown in FIG. By performing the above processing on a large number of pieces of reputation information, it is possible to determine a score for each array stored in the score storage unit 163 shown in FIG.

Note that each of S161 to S16 shown in FIG.
8 is not limited to this, and the execution order of FIG.
The process of S164 may be performed first as described above. Further, S164 may be between S161 and S162 or between S162 and S163.

As described above, according to the present embodiment, the reputation information extracted by the score determination reputation information discriminating means 143 is used by using the score input by the user.
Since the score used for determining the priority order is determined, it is possible to more accurately assign the priority order to the reputation information.

[0083]

The first effect of the present invention is that the automatic determination of the appearance position of the reputation information and the automatic determination of the priority order of the reputation information can be presented to the user in order from the more reputation information. It is. The reason for this is that reputation information discriminating means for extracting reputation information relating to search words contained in the input document, and reputation information priority judgment for assigning priorities to reputation information extracted by the discrimination information discriminating means in order of reputation information likeness This is because it has means.

A second effect of the present invention is that the accuracy of reputation information extraction and the speed of extraction processing are improved. The reason is,
This is because a field in which the search word is used is specified, and evaluation expressions not used in the field are excluded and reputation information is extracted.

A third effect of the present invention is that when prioritizing reputation information, priorities can be assigned more accurately. The reason is that the score input from the user for each reputation information is used to determine the score to be used when determining the priority.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of a first embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a reputation information determination unit 13;

FIG. 3 is a flowchart showing a processing example of reputation information discriminating means 13;

FIG. 4 is a block diagram showing a configuration example of a reputation information priority determination unit 16;

FIG. 5 is a flowchart showing a processing example of a reputation information priority determining means 16;

FIG. 6 is a diagram showing an example of the contents of a rule storage unit 161;

FIG. 7 is a diagram showing an example of contents of a score storage unit 163.

FIG. 8 is a diagram illustrating an example of contents of a determination result storage unit 14;

FIG. 9 is a diagram showing an example of contents of a reputation information storage unit 15;

FIG. 10 is a diagram illustrating an example of an array of reputation information and scores.

FIG. 11 is a diagram showing an example of a priority order of each piece of reputation information.

FIG. 12 is a block diagram illustrating a configuration example of a reputation information determining unit 13a used in a second embodiment of the present invention.

FIG. 13 is a flowchart showing a processing example of the reputation information discriminating means 13a.

FIG. 14 is a block diagram illustrating a configuration example of a score determination device.

FIG. 15 is a block diagram illustrating a configuration example of a score determination unit 146.

FIG. 16 is a flowchart showing a processing example of the score determination device.

FIG. 17 is a diagram illustrating an example of the contents of a score determination determination result storage unit 144;

FIG. 18 is a diagram showing an example of the contents of a score determination reputation information storage unit 145.

FIG. 19 is a diagram showing a display example of reputation information and an input field by the display control means 206.

FIG. 20 is a diagram showing an example of a score input by a user.

FIG. 21 is a diagram showing final contents of a score-with-frequency storage unit 203;

FIG. 22 is a flowchart showing another processing example of the score determination device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Information search device 11 ... Document input means 12 ... Search term input means 13 ... Reputation information discriminating means 131 ... Evaluation expression storage part 132 ... Search term position specifying means 133 ... Evaluation expression position specifying means 134 ... Inter-expression distance judgment means 135 ... reputation information extraction means 14 ... discrimination result storage part 15 ... reputation information storage part 16 ... reputation information priority determination means 161 ... rule storage part 162 ... rule collation means 163 ... score storage part 164 ... score search means 165 ... alignment means 166 ... Presentation means 17 ... Priority order storage unit 2 ... Input device 3 ... Display device K1 ... Recording medium 13a ... Reputation information discriminating means 131a ... Evaluation expression storage unit 133a ... Evaluation expression position specifying means 136 ... Search term field specifying means 141 ... Score Document input means for determination 142 ... Search word input means for score determination 143 ... Reputation information discriminator for score determination Stage 144: Score determination determination result storage unit 145: Score determination reputation information storage unit 146 ... Score determination means 147 ... Display device 148 ... Input device 201 ... Score determination rule storage unit 202 ... Score determination rule matching unit 203 ... Score storage unit with frequency 204: score search unit with frequency 205: score calculation unit 206: display control unit 207: score input unit

Claims (18)

[Claims] 1. Reputation information relating to a designated search word existing in an input document is determined based on an appearance position of the search word in the document and an appearance position of an evaluation expression that determines a value for things. Reputation information discriminating means to be extracted, and reputation information extracted by the reputation information discriminating means,
An information search device, comprising: reputation information priority determining means for assigning priorities in the order of reputation information. 2. The information retrieval apparatus according to claim 1, wherein said reputation information discriminating means includes: a search word position specifying means for specifying a search word appearance position where said search word appears in said document; Evaluation expression storage unit, an evaluation expression position specifying unit that specifies an evaluation expression appearance position where the evaluation expression stored in the evaluation expression storage unit appears in the document, and a search specified by the search word position specification unit. Based on the word appearance position and the evaluation expression appearance position specified by the evaluation expression position specifying means, an inter-expression distance determining means for obtaining a pair of the search word and the evaluation expression within a certain distance from each other; Reputation information extracting means for extracting, from the document, reputation information including at least a portion sandwiched between the search term and the evaluation expression constituting the pair determined by the distance determining means. An information retrieval device characterized by the following. 3. The information retrieval apparatus according to claim 1, wherein said reputation information discriminating means comprises: a search word field specifying means for specifying a field in which said search word is used; and a search wherein said search word appears in said document. Search word position specifying means for specifying a word appearance position; an evaluation expression storage unit storing a plurality of evaluation expressions and field information indicating a field of use of each evaluation expression; and storing in the evaluation expression storage unit in the document Evaluation expression position specifying means for specifying an evaluation expression appearance position at which an evaluation expression used in the field specified by the search word field specifying means, among the evaluation expressions, Expression-to-expression distance determination for finding a pair of a search word and an evaluation expression within a certain range of each other based on the identified search expression appearance position and the evaluation expression appearance position identified by the evaluation expression position identification means. Means, and reputation information extracting means for extracting, from the document, reputation information including at least a portion sandwiched between the search term and the evaluation expression constituting the pair determined by the inter-expression distance determining means. Information retrieval device. 4. The information retrieval device according to claim 1, wherein said reputation information priority determining means is used for determining a priority of each reputation information extracted by said reputation information determining means. A rule group consisting of a plurality of rules, a rule regarding a positional relationship between a search word in reputation information and an evaluation expression, a rule regarding a position of a search word in reputation information, or a rule regarding a position of an evaluation expression in reputation information. A rule storage unit in which a rule group including at least one of the rules is stored; reputation information extracted by the reputation information determination unit; and a plurality of rules stored in the rule storage unit. A rule matching unit that outputs a value indicating whether or not the reputation information satisfies the rule for each rule; and a possibility that the rule matching unit outputs the value. For each combination of values indicating whether or not the rule for each rule is satisfied, a score storage unit in which a score is stored, and a combination of values indicating whether or not the rule output from the rule matching unit is satisfied. For the corresponding score, score search means for searching from the score storage unit, for each reputation information extracted by the reputation information determination means,
An information search device, comprising: a sorting unit that assigns a priority order based on the score of each piece of reputation information searched by the score search unit. 5. The information search device according to claim 4, wherein the score storage unit stores a combination of values that are likely to be output from the rule matching unit in an array format, and the array includes An information search device characterized in that a score is stored in association with a score. 6. A score determination device for determining a score to be stored in a score storage unit in the information search device according to claim 4, wherein reputation information on a designated search word present in an input document is obtained. Score reputation information discriminating means for extracting based on the appearance position of the search word in the document and the appearance position of the evaluation expression that determines the value for things; and each reputation extracted by the score reputation information discriminating means. Score determining means for displaying information on a display device and presenting it to a user, and determining a score to be stored in the score storage unit based on a score for each of the reputation information input by the user using an input device. A score determination device comprising: 7. The score determination device according to claim 6, wherein the score determination means displays each reputation information extracted by the score determination reputation information determination means on the display device, and the input device. A score input unit that captures a score for each piece of reputation information displayed on the display device, input by the user, and a priority order for each piece of reputation information extracted by the score determination reputation information determination unit. A rule group consisting of a plurality of rules to be used, the rule relating to the positional relationship between the search word in the reputation information and the evaluation expression, the rule relating to the position of the search word in the reputation information, or the rule relating to the position of the evaluation expression in the reputation information. A score determining rule storage unit storing a rule group including at least one of the rules; The obtained reputation information is compared with a plurality of rules stored in the score determination rule storage unit, and for each of the rules, a value indicating whether the reputation information satisfies the rule is output. A rule summarizing means for score determination, and a total sum of scores and a total sum of scores for each combination of values indicating whether or not the rule for each rule is satisfied, which may be output from the rule matching means for score determination. A frequency-added score storage unit in which the number of inputs is stored, and a total sum of scores and a total number of input scores corresponding to a combination of values indicating whether or not a rule output from the score determining rule matching unit is satisfied. A frequency score search means for searching from the frequency score storage section, and the sum of the scores searched by the frequency score search means, After adding the score received by the input means and adding 1 to the total number of times of input of the score searched by the score search means with frequency, dividing the total sum of the updated score by the total number of input times of the updated score And a score calculating unit that replaces the total sum of the searched scores and the total number of input scores with the total sum of the updated scores and the total number of input scores of the updated scores. Score determination device. 8. Reputation information on a designated search word present in an input document is determined based on an appearance position of the search word in the document and an appearance position of an evaluation expression that determines a value for things. Reputation information discriminating step to be extracted, and for each reputation information extracted by the reputation information discriminating means,
A reputation information priority determining step of prioritizing in order of reputation information likeness. 9. The information search method according to claim 8, wherein the reputation information determining step includes: a search word position specifying step of specifying a search word appearance position where the search word appears in the document; and an evaluation expression in the document. An evaluation expression position specifying step of specifying an evaluation expression appearance position in which a search term appears, and a search word appearance position specified by the search word position specification step and an evaluation expression appearance position specified by the evaluation expression position specification step. Distance determining step for obtaining a pair of a search word and an evaluation expression whose distance is within a certain range; and a part sandwiched between the search word and the evaluation expression forming the pair obtained in the distance expression determining step. A reputation information extraction step of extracting at least reputation information from the document. 10. The information search method according to claim 8, wherein the reputation information determination step includes: a search word field specifying step for specifying a field in which the search word is used; and a search in which the search word appears in the document. A search word position specifying step of specifying a word appearance position; and a plurality of evaluation expressions and field information indicating a field of use of each of the evaluation expressions. An evaluation expression position specifying step for specifying an evaluation expression appearance position where an evaluation expression used in the field specified in the search word field specifying step appears in the document; and a search word appearance position specified by the search word position specifying step. Based on the evaluation expression appearance position specified in the evaluation expression position specifying step, a distance between expressions for obtaining a pair of a search word and an evaluation expression within a certain distance from each other. A separation determination step; and a reputation information extraction step of extracting, from the document, reputation information including at least a portion sandwiched between the search term and the evaluation expression constituting the pair obtained in the expression distance determination step. Information retrieval method to be characterized. 11. The information search method according to claim 8, 9 or 10, wherein the reputation information priority determination step is performed by the reputation information extracted in the reputation information determination step and the reputation information extracted in the reputation information determination step. A rule group consisting of a plurality of rules used to determine the priority of each reputation information, which is a rule regarding a positional relationship between a search word in the reputation information and an evaluation expression, and a rule regarding a position of a search word in the reputation information. Alternatively, each rule is compared with each rule in a rule storage unit in which a rule group including at least one of the rules regarding the position of the evaluation expression in the reputation information is stored. A rule matching step of outputting a value indicating whether or not the rule is satisfied; and each of the rules, which may be output from the rule matching step. For each combination of values indicating whether or not a rule is satisfied, a score corresponding to a combination of values indicating whether or not the rule output from the rule matching step is satisfied is stored in a score storage unit in which a score is stored. And a sorting step of assigning a priority to each of the reputation information extracted in the reputation information determination step based on the score of each of the reputation information searched in the score search step. An information retrieval method characterized by the following. 12. The information search method according to claim 11, wherein the score storage unit stores combinations of values that may be output from the rule matching step in an array format, and the array includes An information retrieval method characterized in that a score is stored in association with a score. 13. A score determination method for determining a score to be stored in a score storage unit in the information search method according to claim 11, wherein reputation information on an indicated search word existing in an input document is obtained. A reputation information determining step for score extraction, which is extracted based on an appearance position of the search word in the document and an appearance position of an evaluation expression that determines a value for a thing; and each reputation extracted in the reputation information determination step for score determination. Displaying information on a display device and presenting it to a user, based on a score for each of the reputation information input by the user using an input device, a score determining step of determining a score to be stored in the score storage unit; And a score determining method. 14. The score determination method according to claim 13, wherein the score determination step comprises: a display control step of displaying each reputation information extracted by the score determination reputation information determination step on the display device; and the input device. A score input step of taking in a score for each piece of reputation information displayed on the display device from the user, a reputation information extracted in the score determination reputation information determination step, and a score determination reputation information determination A rule group consisting of a plurality of rules used to determine the priority of each reputation information extracted in the step, a rule regarding a positional relationship between a search word in the reputation information and an evaluation expression, a search in the reputation information At least one of the rules regarding the position of a word or the rule regarding the position of an evaluation expression in reputation information
And a rule determination unit that collates each rule in a score determination rule storage unit in which a rule group including a rule is stored and outputs a value indicating whether or not the reputation information satisfies the rule for each of the rules. A rule summing step, and a sum of scores corresponding to a combination of values indicating whether or not the rule output from the score determining rule matching step is satisfied, and a total number of times of inputting the score, May be output from a step,
A frequency score search step for searching from a frequency score storage unit in which a total of scores and a total number of input scores are stored for each combination of values indicating whether or not a rule for each rule is satisfied; After adding the score received in the score input step to the sum of the scores searched in the scored search step and adding 1 to the total number of times of the score searched in the scored search step with frequency, the updated score Is obtained by dividing the total sum of the scores by the total number of input times of the updated score.Furthermore, the total sum of the searched scores and the total number of input times of the score are obtained by the total sum of the updated score and the total number of input times of the updated score. A score calculating step to be replaced. 15. A computer, using reputation information relating to a designated search word present in an input document, by recognizing the appearance position of the search word in the document and the appearance position of an evaluation expression that determines a value for things. Reputation information discriminating means for extracting based on the reputation information, for each reputation information extracted by the reputation information discriminating means,
A program recording medium having recorded thereon a program for functioning as reputation information priority determining means for assigning priorities in the order of reputation information. 16. The reputation information determining means, a search word position specifying means for specifying a search word appearance position where the search word appears in the document, and an evaluation for specifying an evaluation expression appearance position where an evaluation expression appears in the document. An expression position identification unit, based on the search word appearance position identified by the search word position identification unit and the evaluation expression appearance position identified by the evaluation expression position identification unit, a search word whose evaluation distance is within a certain range, An expression distance determination means for obtaining a pair with an expression, reputation information extraction for extracting reputation information including at least a portion sandwiched between a search word and an evaluation expression constituting the pair determined by the expression distance determination means from the document 16. The program recording medium according to claim 15, wherein a program for functioning as means is recorded. 17. The reputation information discriminating means, a search word field specifying means for specifying a field in which the search word is used, a search word position specifying means for specifying a search word appearance position where the search word appears in the document A plurality of evaluation expressions and field information indicating a field of use of each of the evaluation expressions are stored in the evaluation expression storage unit. Evaluation expression position specifying means for specifying an evaluation expression appearance position where the evaluation expression to be appeared in the document, a search word appearance position specified by the search word position specification means, and an evaluation expression appearance position specified by the evaluation expression position specification means , An expression distance determining means for obtaining a pair of a search word and an evaluation expression having a distance from each other within a certain range, and a search word and an evaluation constituting a pair obtained by the expression distance determining means Program recording medium according to claim 15, wherein the recording a program for functioning at least includes reputation information portion sandwiched by the current as reputation information extracting means for extracting from the document. 18. A plurality of reputation information priority determining means for determining a reputation information extracted by the reputation information determining means and a priority order of each reputation information extracted by the reputation information determining means. A rule group consisting of the following rules: a rule regarding the positional relationship between the search word in the reputation information and the evaluation expression, a rule regarding the position of the search word in the reputation information, or a rule regarding the position of the evaluation expression in the reputation information. A rule matching unit that matches each rule in a rule storage unit storing a rule group including at least one of the following, and outputs a value indicating whether or not the reputation information satisfies the rule for each rule: A score in which a score is stored for each combination of values that may be output from the rule matching unit and that indicates whether or not the rule for each rule is satisfied. From 憶部 score search means for searching a score corresponding to the combination of a value indicating whether to satisfy the rule output from said rule matching means for each evaluation information extracted by the reputation information determining means,
2. A program for recording a program for functioning as a sorting unit for assigning a priority based on a score of each of the reputation information searched by the score searching unit.
18. The program recording medium according to 5, 16, or 17.