JP3847977B2 - Information search method, information search device, and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an information collection method for retrieving and collecting unknown information resources related to a set of information resources of interest with respect to information resources such as documents and images that can be referred to by a plurality of users, and information retrieval using the information collection method. With regard to devices, for example, by collaborative filtering technology for acquiring unknown information related to people with similar tastes for content such as movies and music, and the publication and sharing of bookmark functions of WWW (World Wide Web) browsers Involved in technology for promoting information sharing within a group (community) on a network.
[0002]
[Prior art]
The personal information classification or organization method using labeling already exists as an application similar to the conventional memo / annotation function. A feature of these methods is that information is classified and searched using the name of the label attached by the user and the attribute value attached to the label. At this time, if the range of use is limited to individuals, it will not be a problem, but considering that the user can directly use the label name or memo contents that the user has given directly to share information and search with others, The consistency and consistency of the words used in label names and memos is a problem.
[0003]
To avoid this, when building a system, choose whether to unify label names and terms to limit the name space, or to handle name association by the system using a thesaurus or word database. There are many.
[0004]
In the latter case, the problem of word ambiguity and ambiguity remains even if a large-scale dictionary is prepared to improve the search accuracy. In addition, analysis of document text, extraction of keywords, and use for information retrieval are often used when retrieving web documents.
[0005]
However, just extracting a word in a sentence reveals that the word exists, but it is difficult to analyze what kind of viewpoint the document is written in and useful. In addition, keyword analysis is not possible for information resources with only image, audio and video information. In any case, the evaluation and classification of how useful the document is for the user and from what point of view the document is useful for each user are different for each user.
[0006]
Some systems actively use evaluation information performed by users. A typical example is a system using collaborative filtering called a recommendation engine (Recommendation Engine). This is one method for accurately presenting a list that suits the user's preference when advertising and selling videos, music, and books on the Internet. Specifically, for example, by making a database of numerical evaluations (like or dislike or how much you like) given by many users to many movies, and calculating the correlation of each person's preference, A list of movie titles that the user is likely to like is presented from a group similar to the user's preferences. A representative example of the recommendation engine is a GroupLens recommendation engine (USA, Net Perception). The current recommendation engine can be regarded as giving a standard label for evaluation (for example, a mark sheet-like evaluation label in which “like” to “dislike” are quantified) to data such as movies and music.
However, the reality is that content can only be evaluated in one way, such as whether you like it or not.
[0007]
[Problems to be solved by the invention]
Thus, in a search engine based on normal keywords such as “Yahoo” and “Infoseek” on the WWW, the search and the act of collecting and organizing information (bookmarks and filing) are completely separated. Therefore, even if information is collected about a certain matter, there is a problem that it cannot be used to improve the accuracy of search.
[0008]
Therefore, the present invention uses information obtained as a result of collecting and classifying an arbitrary information resource such as a document or an image that can be referred to by a plurality of users on a network based on a free viewpoint, It is an object of the present invention to provide an information search method and an information search apparatus using the same, which can efficiently search and collect unknown information resources related to information of interest.
[0009]
[Means for Solving the Problems]
(1) An information search method according to the present invention (Claim 1) is one of a plurality of labeling sets composed of a plurality of information resources characterized by a correspondence relationship between information resources and labels assigned to the information resources. One or more search information resources (stored in the collector) characterized by the logical expression of the label selected from (for example, labeling set A) are replaced with one or more other labeling sets (for example, labeling set). Search from each of B, C)
Obtaining a first evaluation value (label score) for evaluating a link between the search information resource and the label for the label attached to the searched information resource;
Based on the first evaluation value, a second evaluation value (information resource score) that evaluates the degree of association of the information resource other than the search information resource with the search information resource is obtained.
An information resource related to the information resource for search selected based on the second evaluation value is presented.
[0010]
According to the present invention, it is possible to efficiently search for an unknown information resource related to information of interest using a result obtained by organizing and classifying information resources collected by each user based on a free viewpoint by labeling (labeling). Can be collected.
[0011]
(Claim 2) Preferably, the second evaluation value of the information resource obtained from each of the plurality of labeling sets is integrated based on a predetermined weighting factor for each of the plurality of labeling sets, and the integration result The information resource related to the information resource for search is selected based on the above.
[0012]
(Claim 3) Preferably, for a label attached to the search information resource, a predetermined value is cumulatively added for each search information resource to which the label is attached, so that the first evaluation of the label is performed. Calculate the value.
[0013]
(Claim 4) Preferably, for the label attached to the search information resource, a first evaluation value of the label is calculated based on the importance of the search information resource to which the label is attached.
[0014]
(Claim 5) Preferably, for an information resource other than the information resource for search, a second evaluation of the information resource is performed by cumulatively adding the first evaluation value of a label attached to the information resource. Calculate the value.
[0015]
(Claim 6) Preferably, for an information resource other than the information resource for search, a second of the information resource is determined based on the first evaluation value of the label attached to the information resource and the importance of the label. The evaluation value of is calculated.
[0016]
(2) The information search device of the present invention (Claim 7) is one of a plurality of labeling sets including a plurality of information resources characterized by a correspondence relationship between information resources and labels assigned to the information resources. Search means for searching one or more search information resources characterized by a logical expression of the label selected from each of the other one or more labeling sets;
Label evaluation means for obtaining a first evaluation value for evaluating a link between the search information resource and the label for the label attached to the search information resource searched by the search means;
Information resource evaluation means for obtaining a second evaluation value for evaluating the degree of association of the information resource other than the information resource for search with the information resource for search based on the first evaluation value;
Presenting means for presenting an information resource related to the information resource for search selected based on the second evaluation value;
It is characterized by comprising.
[0017]
According to the present invention, it is possible to efficiently search for an unknown information resource related to information of interest using a result obtained by organizing and classifying information resources collected by each user based on a free viewpoint by labeling (labeling). Can be collected.
[0018]
(Claim 8) Preferably, the second evaluation value of the information resource obtained from each of the plurality of labeling sets is integrated based on a weighting factor predetermined for each of the plurality of labeling sets, and the integration result The information resource related to the information resource for search is selected based on the above.
[0019]
(Claim 9) Preferably, for a label attached to the search information resource, a predetermined value is cumulatively added for each search information resource to which the label is attached, so that the first evaluation of the label is performed. Calculate the value.
[0020]
(Claim 10) Preferably, for the label attached to the search information resource, a first evaluation value of the label is calculated based on the importance of the search information resource to which the label is attached.
[0021]
(Claim 11) Preferably, for an information resource other than the information resource for search, a second evaluation of the information resource is performed by cumulatively adding the first evaluation value of a label attached to the information resource. Calculate the value.
[0022]
(Claim 12) Preferably, for information resources other than the information resource for search, based on the first evaluation value of the label attached to the information resource and the importance of the label, An evaluation value of 2 is calculated.
[0023]
(3) The recording medium of the present invention (Claim 13) is provided by one of a plurality of labeling sets including a plurality of information resources characterized by a correspondence relationship between information resources and labels assigned to the information resources. Search means for retrieving one or more search information resources characterized by the logical expression of the selected label from each of the other one or more labeling sets;
Label evaluation means for calculating a first evaluation value for evaluating a link between the search information resource and the label for the label attached to the search information resource searched by the search means;
Information resource evaluation means for calculating a second evaluation value for evaluating a degree of association of the information resource other than the search information resource with the search information resource based on the first evaluation value;
Presenting means for presenting an information resource related to the information resource for search selected based on the second evaluation value;
Is a machine-readable recording medium recording a program for executing the above, and according to the present invention, for example, a screen (pallet) for performing an operation on a label, and a collector used for information retrieval are created. And a display device for presenting a user interface screen such as a finder for presenting search results, and a storage device such as a hard disk device (preferably further capable of communication via a network such as the Internet) As programs that can be executed by a computer such as a personal computer, each program is stored and distributed in a recording medium such as a magnetic disk (floppy disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc. Collect based on user's free viewpoint Labeling the information resource (labeled) using the results of the organizing classification enables efficient search and collect unknown information resources associated with information of interest.
[0024]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
Here, terms used in the following description are defined.
The information resource is, for example, an information file such as voice, text, still image, moving image, HTML document, or the like.
[0025]
The act of giving a label with a subjective free name based on an arbitrary standard or viewpoint to multiple information resources is called `` labeling '', and the set of correspondences between labels and information resources obtained as a result of labeling This will be called “labeling set”.
(First embodiment)
FIG. 1 schematically shows a basic configuration of an information search apparatus using an information search method according to the first embodiment of the present invention. The information search processing unit 1 and data storage are roughly divided. The information search processing unit 1 further includes a search information resource list acquisition unit 10, a labeling set selection unit 20, an information search unit 30, and a candidate information resource presentation unit 40. Yes.
[0026]
Here, for example, for information resources collected from the Internet using a browser or created by the user, labels that the user decides freely based on criteria and viewpoints that the user wants to organize or record are attached. It is assumed that information about the label, information about the labeled information resource, and the like are already stored in the data storage unit 50.
[0027]
FIG. 2 conceptually shows two labeling sets A and B. Here, for example, the labeling set A and the labeling set B indicate labeling sets possessed by different users, but one user may have a plurality of labeling sets A and B.
[0028]
One labeling set includes a set of labels defined therein and a set of arbitrary information resources to which the labels are attached. For example, in the labeling set A, there is a label set {A1, A2, A3} and a set of information resources {a, b, c, d, e} to which those labels are attached.
[0029]
In FIG. 2, the relationship between the label and the information resource, that is, the relationship that a certain label is attached to a certain information resource is expressed by a line connecting the label and the information resource.
[0030]
In the labeling set B, similarly to the labeling set A, there exists a label set {B1, B2, B3, B4}, and these labels are attached to the information resources {a, b, e, f, g, k}. It has been. It shows that information resources with the same name refer to the same entity. At this time, for example, in the labeling set A, the set of information resources having both the label A1 and the label A2 may be considered to include matters relating to both the label A1 and the label A2.
[0031]
As described above, as a result of labeling, information resources are classified by various cuts according to combinations of label logical expressions. The logical expression of the label is expressed in the form of “A1 and A2”, for example, and R (A1 and A2) means a set of information resources with both the label A1 and the label A2. For example, in the labeling sets A and B shown in FIG. 2, R (A1 and A2) = {a, b}.
[0032]
The main object of the present invention is to search and collect unknown information resources related to those information resources from another labeling set using a set of information resources characterized by such a logical expression of labels as a search key. That is.
[0033]
The search information resource list acquisition unit 10 of FIG. 1 acquires a list of information resources expressed by a logical expression of labels in a labeling set, and uses this as a search key. This process corresponds to obtaining the information resource list {a, b} by designating the label logical expression “A1 and A2” as described above in the labeling set A of FIG. 2, for example. Basically, the information resource list {a, b} is obtained by a logical expression of a label, but a specific information resource can be removed from or added to the information resource list as necessary.
[0034]
After obtaining an information resource list (specifically, an information resource identification information (ID) list, also referred to as a main list) that represents a set of information resources for search, Decide the target (other labeling set) to search for related information. Therefore, the labeling set selection unit 20 selects another labeling set to be searched, and notifies the partner labeling set of the information resource list obtained by the search information resource list acquisition unit 10. That is, if it is in-house as a search target, a specific department is specified and limited to a labeling set possessed by a person belonging to the department, or a specific person is designated and limited to a labeling set possessed by the person Limit the search range. If this is not necessary, it is not always necessary to limit the search range.
[0035]
In the example of FIG. 2, this corresponds to selecting the labeling set B and notifying the labeling set B of the information resource list {a, b}. In general, there are a plurality of labeling sets to be selected, but all labeling sets other than the labeling set (here, labeling set A) that is the source of the search implicitly become search targets.
[0036]
Next, the information search unit 30 searches related information for each labeling set.
The information search unit 30 includes a label score calculation unit 31, an information resource score calculation unit 32, and a candidate information resource list creation unit 33.
[0037]
Here, the label score represents a degree of association between a set of information resources Rs given for search and a label defined in a certain labeling set. The information resource score is an index for measuring the relationship between Rs and other information resources calculated based on the label score. Briefly, an information resource that is labeled the same as the given information resource set Rs has a higher score, and the information resource that is highly related to the original Rs in terms of labeling. You can say that.
[0038]
FIG. 3 is a flowchart showing the processing operations of the label score calculation unit 31 and the information resource score calculation unit 32, and FIG. 4 illustrates the processing (label score calculation processing) from step S2 to step S4 in FIG. FIG. 5 is a conceptual diagram for explaining the processing (information resource score calculation processing) from step S5 to step S7 in FIG.
[0039]
First, the processing operation (label score calculation processing) of the label score calculation unit 31 will be described with reference to FIGS.
In step S1 of FIG. 4, the label score and the information resource score are initialized to a value of “0”, for example. Next, in step S2, when the information resource set for search is Rs, it is checked whether all elements (information resources) of Rs have been examined. If elements to be examined still remain, step S3 is performed. Go to, extract one element from Rs and let it be r. In step S4, a predetermined numerical value (for example, “1”) is added to the label scores of all labels attached to the extracted information resource r.
[0040]
In FIG. 4, when an information resource {a, b} is given as Rs, in the labeling set B, the label attached to each element (information resource) of Rs is attached 1 For each information resource, the numerical value “1” is added to the label score equally. The label B1 is the label of the information resource a, b given as Rs, the label B2 is the label of the information resource b given as Rs, and the label B4 is the label of the information resource b given as Rs. Therefore, the label scores Sc (B1), Sc (B2), and Sc (B4) of the labels B1, B2, and B4 are Sc (B1) = 1 + 1 = 2, Sc (B2) = 1, and Sc (B4), respectively. ) = 1.
[0041]
Here, the numerical values to be added per information resource for all labels are all equal to “1” because the element (information resource) of the information resource set Rs selected from the labeling set A This is because it is considered appropriate to uniformly add the same numerical value to the labels attached to the elements of Rs on the labeling set B when no ranking such as importance exists between them.
[0042]
If there is a difference in importance between the label A1 and the label A2 on the labeling set A, for example, Rs = R (A1 or A2) = {a, b as a set of information resources for searching from the labeling set A , D, e} ("or" is a logical operation symbol meaning "or", and in this case means a set of information resources to which at least one of label A1 or label A2 is attached) At this time, the importance can be assigned to these four information resources according to the importance of the assigned label. Therefore, when calculating the label score on the labeling set B, each information resource is assigned a numerical value with a high or low value according to this importance, and the value assigned to the label attached to the information resource. May be added.
[0043]
Next, the processing operation (information resource score calculation process) of the information resource score calculation unit 32 will be described with reference to FIGS.
In the information resource score calculation unit 32, a value obtained by adding the label score value of each label obtained by the label score calculation unit 31 for information resources other than Rs to which the label is attached is set as an information resource score. Here, a set of labels attached to a set Rs of information resources for search on the labeling set B is L (Rs).
[0044]
In step S5 in FIG. 3, it is checked whether all the elements (labels) of L (Rs) have been examined. If there are still labels to be examined, the process proceeds to step S6, and one element is selected from L (Rs). Take it out and let it be λ. In step S7, the label score of λ is added to the score of the information resource other than Rs to which the extracted label λ is attached.
[0045]
In FIG. 5, the label score values of the labels B1, B2, and B4 obtained by the label score calculation unit 31 are used as information resources {m, g} other than Rs to which the labels are attached.
Add about. For example, the information resource m has a label B1 and a label B2. The score Sc (m) of the information resource m is obtained by adding the label scores Sc (B1) and Sc (B2) of the label B1 and the label B2 attached thereto. That is, Sc (m) = Sc (B1) + Sc (B2) = 2 + 1 = 3. Similarly, regarding the score of the information resource g, Sc (g) = Sc (B4) = 1 from the label score of the label B4.
[0046]
At this time, if the importance is given to the label, the contribution from each label score can be changed depending on the importance when the information resource score is obtained. For example, a numerical value (for example, “2” is assigned to a higher importance level and “1” is assigned to a lower importance level) assigned to each label according to the importance level. You may make it add, after multiplying with the score of a resource.
For example, if “2” is assigned because label B1 has high importance and “1” is assigned because label B2 has low importance, the score of information resource m is Sc (m) = {Sc (B1) · 2} + {Sc (B2) · 1} = 4 + 1 = 5.
[0047]
The candidate information resource list creation unit 33 selects an information resource having a high information resource score value as a candidate information resource list (candidate list), and creates a list in which the information resources are ranked based on the score value. For example, in the example of FIG. 5, the only information resources whose information resource score is not “0” are m and g, and therefore, a set of candidate information resources Rc (B) = {m, g} retrieved from the labeling set B And the first candidate information resource is m, and the second candidate information resource is g, depending on the value of each information resource score.
[0048]
Finally, the candidate information resource presentation unit 40 presents the ranked list of candidate information resources created by the candidate information resource list creation unit 33 to the original labeling set that issued the search.
[0049]
The above has been described assuming that one labeling set is the target of the search. However, in the case of information search across a plurality of labeling sets, basically, the result of information search obtained from the labeling set of each search target Are superimposed (integrated).
[0050]
FIG. 6 conceptually shows a case where information resources are searched from a plurality (for example, four) of labeling sets A to D.
For example, when a search is issued from the labeling set A as described above, the candidate information resource sets Rc (B), Rc (C), and Rc (D) are obtained in the same manner as described above for each of the labeling sets B to D. . At this time, the logical sum of these sets of candidate information resources is calculated, and the information score for the same candidate information resource is added to each labeling set multiplied by a weighting factor given in advance to obtain an integrated score. .
[0051]
Specifically, for example, Rc (B) = {m (3), g (1)}, Rc (C) = {g (2) as candidate information resources from the labeling sets B, C, and D, respectively. , M (1), p (1)}, Rc (D) = {m (2), p (2), e (1)}. However, the numerical value in parentheses of each element (information resource) represents the value of the information resource score. At this time, the logical sum of these candidate information resources is calculated, and the final set of candidate information resources is Rc = {m, g, p, e}. Assuming that KB, KC, and KD are given as the weighting coefficients of the labeling sets B, C, and D, the integrated score of the candidate information resource m is the information resource score in each of the labeling sets B, C, and D. Since it is sufficient to add after multiplying the weighting factor,
Sc (m) = 3 · KB + 1 · KC + 2 · KD
It becomes. Normally, it can be considered that the weighting factors of each labeling set are all the same, and in this case, the integrated score of each candidate information resource may be a simple sum of the information resource scores calculated in each labeling set. Become.
(Second Embodiment)
FIG. 7 conceptually shows a case where information is retrieved and collected using the information retrieval method of the present invention. The user 80 attaches a label defined by the user to the information resource. If you want to search for information related to a set of information resources, first create a virtual folder called collector 81, which will be described later, using the logical expression of the label characterizing that information resource set as a search key. The corresponding information resource is extracted from the labeling set. Next, a search is performed on a plurality of other labeling sets 83 using the information resource list as a search key. Search results 84 are obtained as a ranked list of candidate information resources. It is possible to add some candidate information resources to the collector as needed.
[0052]
FIG. 8 conceptually shows a case where information search is performed via a network, for example, using the information search apparatus according to the second embodiment of the present invention. The information resource is on the intranet and the Internet, and can be accessed by a plurality of users (for example, two users α and β). In FIG. 8, the information resource is stored in the data storage units 402, 422, and 432. Yes. In this example, it is assumed that the subject that performs labeling is a user, that is, a person.
[0053]
For example, the user α uses an information search device (hereinafter, simply referred to as a terminal) 421 as a client to label information accessed by the user α. Information relating to the user, information relating to the label created by the user, information relating to the information resource labeled by the user, information relating to the collector storing the information resource, etc. are stored in the data storage units 422 and 432 of each client (terminals 421 and 431). In addition to being stored, it is sent via the network 400 to a labeling server 411 that manages and processes labeling information, and is also stored in the labeling DB 412. When searching for information between labeling sets from a client, the labeling server 411 is requested to perform necessary processing. The information held and managed by the labeling server 411 is managed so as to be consistent with the data held on each client side. Each user's terminal 421, 431 holds information about the label as a result of each user's own labeling, information about the information resource labeled by the user, information about the collector storing the information resource, etc. Via the labeling interface in the terminals 421 and 431, information is searched for labeling and labeling of yourself and others. The labeling interface is composed of three basic components: pallet, finder, and collector.
[0054]
FIG. 9 schematically shows a configuration of a main part of an information search apparatus (corresponding to the terminals 421 and 431 in FIG. 8) according to the second embodiment of the present invention. The structure of the information retrieval apparatus which concerns on is shown more concretely.
[0055]
The information search apparatus shown in FIG. 9 includes a user interface unit 501, a processing unit 502, a data storage unit 506, and a transmission / reception unit 507.
[0056]
The processing unit 502 includes a labeling processing unit 503 for performing labeling processing, a collector creation processing unit 504 for performing collector creation processing, and an information search processing unit 505 for performing information retrieval processing.
[0057]
The labeling processing unit 503 includes a label definition unit 510 and a label assignment unit 520. The collector creation processing unit 504 includes a collector search formula setting unit 530, a main list acquisition unit 540, and a main list operation unit 550. The information search processing unit 505 includes a collector selection unit 560, an information search unit 570, and a candidate information resource list presentation unit 580.
[0058]
The user interface unit 501 is for presenting and inputting information necessary for executing the labeling process, the collector creation process, and the information search process in each of the above units. For example, an input from a display device, a mouse, a keyboard, or the like It consists of devices.
[0059]
The data storage unit 506 stores information resources collected by the user, label information that is a result of labeling the information resources, and the like. The collector created by the collector creation processing unit 504 is also stored.
[0060]
The transmission / reception unit 507 transmits the collector created by the collector creation processing unit 504 as a search request to the labeling server 411 via the network 400 and receives the search result via the network 400.
[0061]
FIG. 10 shows the operation of the information search apparatus of FIG. 9 by displaying a user interface screen (pallet 610 for manipulating labels and a list of information resources related to each label and collector) presented in the user interface unit 590. This is conceptually shown based on the relationship between the finder 620) and the collector 630 used mainly for information retrieval.
[0062]
As shown in FIG. 10, in the pallet 610, a label is operated, and in the finder 620, a list of information resources related to each label and collector is displayed, and a collector 630 used for information search is created, and the search result Will be presented.
[0063]
First, the labeling processing unit 503 in FIG. 9 will be described.
Labeling, that is, a process of assigning a user-defined label to a certain information resource is executed by the label definition unit 510 and the label addition unit 520. These processes are performed by presenting an interface screen such as a palette 610 to the user interface unit 501.
[0064]
In the label definition unit 510, the user freely determines the name of the label and creates a new label. If necessary, change the label name or delete the label. The label assignment unit 510 assigns a label to the information resource.
[0065]
The palette 610 acquires the ID (for example, URL) of the currently displayed information resource via an interface screen called another browser 600, and provides a label to be given to the acquired information resource by a mouse operation or the like. It is an interface screen for selecting and labeling.
[0066]
FIG. 11 shows an example of a basic screen of the pallet 610 presented by the user interface unit 501.
As shown in FIG. 11, labels (“cooking”, “health”, “office”, “photograph”, “Hong Kong movie”, and “article”) created by the user are arranged on the basic screen of the palette.
[0067]
When trying to create a new label, if a button 614 “New” is pressed using a mouse or the like, a label input area is presented at an arbitrary position, and a desired label name is input there. do it. Further, when trying to erase a label that has already been created, the label to be erased can be selected (the selected label is highlighted, for example), and the “Delete” button 615 can be pressed. .
[0068]
Since the user can move the label to an arbitrary position by a drag and drop operation, for example, labels similar in genre (for example, “dish” and “health”) can be collected and grouped.
[0069]
When a user wants to attach a label to an information resource on the palette 610, the user inputs a desired information resource ID (for example, URL) in the information resource ID designation area 611, or a downward arrow is displayed. A desired information resource ID is selected from the list of selectable information resource IDs displayed by pressing the button 616. Alternatively, the ID of the information resource opened on the browser 600 may be automatically taken into the information resource ID designation area 611. Next, a desired label is selected from the labels presented on the palette 610 (the selected label is highlighted on the screen), and a label labeling instruction button 612 “Labelling” is displayed. You can press.
[0070]
If you want to change the label once pasted, enter (or select) the ID of the information resource whose label you want to change in the information resource ID designation area 611, and the label already pasted to that information resource will be selected on the screen. Since the state (for example, highlighted on the screen) is displayed, the selection state may be changed and the “Labelling” button 612 may be pressed again.
[0071]
When it is desired to refer to the list of information resources to which the designated label is pasted, for example, when it is desired to refer to the list of information resources to which the label “Cooking” is pasted, the label “Cooking” is displayed on the palette 610 screen. And a button “613” “Finder” may be pressed. Then, a finder 620 to be described later is activated, and the finder 620 presents a list (ID) of information resources to which the label “dishes” is pasted.
[0072]
When a plurality of labels are selected, a list of information resources to which all the selected labels are attached is presented.
As described in the first embodiment, when the importance level is set to the label, it is desirable to set it on the screen of the pallet 610. For example, a desired label is selected on the pallet 610 screen, and a property setting screen for the label is opened. Then, for example, the five levels of importance (“5” when the importance is the highest and “1” when the importance is the lowest) may be set on the property setting screen for each label.
[0073]
In addition to the importance, the label property may include access permission information for limiting the accessible range to the information resource with the label. For example, an information resource with a label that is set as “accessible” as a property can be searched, and an information resource with a label that is set as “not accessible” On the other hand, it is possible to perform an operation such that it cannot be a search target. In addition, as a label property, the labeling set that can be accessed (the information resource can be searched) is limited to the information resource with the label, or conversely, the labeling set that cannot be accessed is specified. You can also do it. As described above, it is desirable to set the property of these labels on the screen by selecting a desired label on the screen of the palette 610 and opening the property setting screen for the label.
[0074]
Next, the collector creation processing unit 504 in FIG. 9 will be described.
In FIG. 14, when a user frequently refers to an information resource labeled with “Hong Kong movies” and “photos” in a labeling set possessed by a certain user, or wants to search for related information, it corresponds. A virtual folder characterized by the logical expression of the label is created, and if it is opened, a list of information resources matching the logical expression of the label is obtained. This virtual folder is called a collector. A logical expression of a label characterizing the collect at that time is called a collector search expression.
[0075]
The collector search formula setting unit 53 in FIG. 9 sets this collector search formula.
In FIG. 14, the collector 710 has a collector search expression of “Hong Kong movie and photo”, and when the collector 710 is opened, it can be seen that information resources {a, c} are contained. At the time of information retrieval, a labeling set (for example, “person”) to be retrieved is designated for each collector, and parameters for retrieval are set.
[0076]
As shown in FIG. 15, when the contents of a certain collector 830 are displayed on the finder 620, the contents of the collector are divided into two. One is a main list 810 and the other is a candidate list 820.
[0077]
The main list 810 is a list of information resources that belong to the collector among the information resources in the labeling set that is a search transmission source, and the candidate list 820 is searched from other labeling sets that are candidates belonging to the collector. A list of information resources. All the information search results from other labeling sets enter this candidate list 820 and can be moved to the main list as needed at the discretion of the user.
[0078]
You can also name collectors. The collector 710 in FIG. 14 is named “Hong Kong movie”, and the collector 631 in FIG. 15 is named “healthy food”.
[0079]
The main list acquisition unit 540 in FIG. 9 creates the main list 810 from the collector search formula set by the collector search formula setting unit 530. The main list operation unit 550 performs an operation of removing a specific information resource from the main list 810 or moving a specific information resource from the candidate list 820 containing the information search result to the main list. .
[0080]
The collector created by the collector creation processing unit 504 is stored in the data storage unit 506.
FIG. 12 shows an example of a basic screen of the finder 620 presented on the user interface unit 501 when creating a collector. For example, by clicking “Collector” in the menu bar 625 of FIG. 12 with a mouse, a collector folder and a collector icon are first created and presented on the basic screen. When the presented icon is double-clicked, the finder 620 is activated and a screen divided into two upper and lower parts as shown in FIG. 12 is presented. The top is a main list presentation area 623 and the bottom is a candidate list presentation area 624. When the collector search formula is input to the collector search formula setting area 621 as a logical expression of the label, the collector search formula setting unit 530 passes it to the main list acquisition unit 540. The main list acquisition unit 540 searches for information resources that match the collector search formula from a local labeling set (for example, a labeling set that has been labeled by the user and stored in the data storage unit 506). A main list that enumerates a set of information resources obtained as a result is created and presented in the main list presentation area 623.
[0081]
In the main list presentation unit 623, the title of the information resource, the ID (for example, URL) of the information resource, and the label attached to the information resource are presented as the main list.
[0082]
Based on the operation performed by the user on the main list presented in the main list presentation area 623, the main list operation unit 550 adds / deletes information resources, and sets the information resources in the main list so as to suit the user's viewpoint. Can be changed.
[0083]
As described in the first embodiment, when the importance level is set for the information resource, it is desirable to set it on the screen of the finder 620. For example, as shown in FIG. 18, an importance setting area 626 for setting the importance of each information resource is provided in the main list presentation area 623. For example, it is only necessary to set a desired numerical value in the importance setting area 626 for each information resource with five levels of importance (“5” when the importance is the highest and “1” when the importance is the lowest).
[0084]
Further, it may be set in advance that information resources displayed at the top of the main list have higher importance. In FIG. 12, the information resource titled “XXX page” has the highest importance, and the information resource titled “health promotion page” has the lowest importance. In this case, for example, statistics on the usage frequency of the information resource as a collector may be collected as needed, and a high importance may be automatically set for an information resource with a high usage frequency. When presenting a collector, the higher the usage frequency, the higher the main list. In addition, the user may set the importance of the information resource by operating the main list presented in the main list presentation area 623 and changing the order of presentation of the information resource in the main list. Such importance level setting processing for information resources may be performed by the main list operation unit 550, for example.
[0085]
Next, the information search processing unit 505 in FIG. 9 will be described.
In the information search processing unit 505, the selected state of the collector as shown in FIG. 12, that is, the setting of the collector search formula and the main list created based on the collector search formula are displayed in the main list presentation area 623. For example, when “Search” in the menu bar 621 is double-clicked with the mouse from the presented state, the collector (at least the main list) is sent to the labeling server 411 on the network 400 via the transmission / reception unit 507. Pass and make a search request. In addition, the information search unit 570 of the information search processing unit 505 executes a search using the collector.
[0086]
When making a search request, as shown in FIG. 14, a search range may be set for the collector.
The labeling server 411 is configured as shown in FIG. 17, for example. In addition, the same code | symbol is attached | subjected to the same part as FIG. In the labeling server 411, the search information resource list acquisition unit 10 receives the collector via the network 400 and the transmission / reception unit 60, and when the search range is set for the collector in the labeling set selection unit 20, Select the target labeling set. In the information search unit 30, a list of candidate information resources (candidate list) is created by performing a label score calculation process, an information resource score calculation process, an integrated score calculation, and the like as shown in FIG. Are sent back to the information retrieval apparatus via the transmission / reception unit 60 and the network 400. In the information search apparatus, as shown in FIG. 13, the candidate list is presented in the candidate list presentation area 624.
[0087]
In this way, by setting a collector search formula for the collector in advance and executing a search from the finder 620, information resources matching the collector search formula can be fetched from the labeling server 411. The acquired information resource is presented separately as a candidate list and stored in the collector, so that the main list customized by the user is not disturbed. The user may transfer only necessary information resources presented in the candidate list presentation area 624 as a candidate list to the main list.
[0088]
The collector selection unit 560 is for selecting a collector for search among the collectors stored in the data storage unit 506 or created by the collector creation processing unit 504.
[0089]
The information search unit 670 has, for example, the same configuration as the information search processing unit 1 in FIG. 1, receives the collector selected by the collector selection unit 560 in the search information resource list acquisition unit 10, and performs the first implementation. As described in the embodiment, information resources are searched from the data storage unit 506 (similar to the processing operations of the search information resource list acquisition unit 10, the labeling set selection unit 20, and the information search unit 30 in FIG. 1).
[0090]
The candidate information resource list presenting unit 580 is for presenting a list of candidate information resources obtained as a result of searching by the information search unit 570.
Next, referring to FIG. 16, the score calculation of candidate information resources is performed by taking as an example a case where information is searched for another two labeling sets B and C using the collector 907 created from the labeling set A. This will be specifically described. It is assumed that the users A, B, and C are classified from the respective viewpoints, and as a result, classifications such as a labeling set A, a labeling set B, and a labeling set C are obtained.
[0091]
For example, in the labeling set A, information resources with a resource ID of dat-03 are labeled with labels A1 and A3, and information resources with a resource ID of dat-10 are labeled with labels A1 and A2. Yes. Hereinafter, similarly, in the notation of the labeling sets A to B in FIG. 16, in order to indicate the label attached to the information resource, a circle is marked at the position where the resource ID and the label intersect.
[0092]
At this time, when the user A generates the collector 907 and gives “A1 and A2” as the collector search expression, the set of information resources matching the collector search expression in his / her own labeling set is {dat -10, dat-36, dat-42}. That is, {dat-10, dat-36, dat-42} is written in the main list 909 of the collector 907.
[0093]
In FIG. 16, the circles marked at the positions where the resource IDs of the information resources that match the collector search expression and the labels attached thereto are crossed out are shown in black.
[0094]
In the labeling set B, information resources that hit (that is, match one of the information resources specified by the collector 907) are dat-10 and dat-36, and dat-10 is labeled B1 and B3. And dat-36 is labeled B3. If all the numerical values to be added for each of these labels are “1”, the label score for label B1 is “1” because only dat-10 is hit, and the label scores for label B3 are dat-10 and dat-36. It is “2” because it is a hit.
[0095]
On the other hand, in the labeling set C, information resources that hit (that is, match any of the information resources specified by the collector 907) are dat-10, dat-36, and dat-42, and dat-10. Are labeled C2, C3, dat-36 is labeled B2, C3, C4, and dat-42 is labeled C2. Assuming that all the numerical values to be added for each of these labels are “1”, the label score of label C2 is “3” because dat-10, dat-36, and dat-42 are hit, and the label score of label C3 is dat−. 10. Since dat-36 is hit, it is “2”, and the label score of label C4 is “1” because only dat-10 is hit.
[0096]
A set of candidate information resources retrieved from the labeling set B using the collector 907 is {dat-20, dat-22, dat-53, dat-60}.
A set of candidate information resources retrieved from the labeling set C using the collector 907 is {dat-04, dat-19, dat-22, dat-60}. In the notation of the labeling sets A and B in FIG. 16, double circles are written at the positions where the resource IDs of the candidate information resources searched from each labeling set and the labels attached thereto intersect.
[0097]
In the labeling set B, information resource scores of the candidate information resources dat-20, dat-22, dat-53, and dat-60 are calculated. The information resource score of the information resource dat-20 is the label score “2” of the label B3, the information resource score of the information resource dat-22 is the label score “1” of the label B1, and the information resource score of the information resource dat-53 is The label score “1” of the label B1 and the label score “2” of the label B3 are added to become “3”, and the information resource score of the information resource dat-60 becomes the label score “1” of the label B1. That is, in the labeling set B
The information resource score of dat-20 is “2”
The information resource score of dat-22 is “1”
The information resource score of dat-53 is “3”
The information resource score of dat-60 is “1”
It is.
[0098]
On the other hand, in the labeling set B, information resource scores of the candidate information resources dat-04, dat-19, dat-22, and dat-60 are calculated. The information resource score of the information resource dat-04 becomes the label score “1” of the label C4, and the information resource score of the information resource dat-19 adds the label score “2” of the label C3 and the label score “1” of the label C4. The information resource score of the information resource dat-22 is “5” by adding the label score “3” of the label C2 and the label score “2” of the label C3, and the information resource dat-60 information The resource score is the label score “3” of the label C2. That is, in the labeling set B,
The information resource score of dat-04 is “1”
The information resource score of dat-19 is “3”
The information resource score of dat-22 is “5”
The information resource score of dat-60 is “3”
It is.
[0099]
Next, an integrated score of candidate information resources retrieved from the labeling sets B and C using the collector 907 is obtained. Here, the weighting coefficients of the labeling sets B and C are both “1”. Then, each integrated score of the information resources dat-20, dat-22, dat-53, dat-60, dat-04, and dat-19 is the information resource score in the labeling set B and the information resource score in the labeling set C. And add
The integrated score of the information resource dat-20 is “2” + “0” = “2”
The integrated score of the information resource dat-22 is “1” + “5” = “6”
The integrated score of the information resource dat-53 is “3” + “0” = “3”
The integrated score of the information resource dat-60 is “1” + “4” = “5”
The integrated score of the information resource dat-04 is “0” + “1” = “1”
The integrated score of the information resource dat-19 is “0” + “3” = “3”
It becomes. A candidate list 910 in which candidate information resources are ranked as shown in FIG. 16 is obtained by rearranging the integrated scores in descending order. The candidate list 910 is stored in the collector 907 (see FIG. 15).
[0100]
As described above, according to the second embodiment, each user collects arbitrary information resources such as documents and images that can be referred to by a plurality of users on the network based on a free viewpoint, By using the classification information obtained as a result of labeling, unknown information resources related to the information of interest can be efficiently searched and collected via the network.
[0101]
Of course, not only information retrieval via a network but also offline information retrieval is possible. That is, the data storage unit 506 of the information search apparatus in FIG. 9 stores information resources collected by the user, information on labels as a result of the user's own labeling on the information resources, information on collectors, and the like. Therefore, if the user has a plurality of labeling sets, one or more search information resources characterized by the logical expression of the label are selected from one of the labeling sets, Based on the labeling set, information resources related to the information resource for search may be searched and presented.
[0102]
【The invention's effect】
As described above, according to the present invention, a user can easily search for information related to collected information while organizing information with a high degree of freedom through subjective labeling.
[Brief description of the drawings]
FIG. 1 is a diagram schematically showing a basic configuration of an information search apparatus using an information search method according to a first embodiment of the present invention.
FIG. 2 is a diagram conceptually illustrating a state in which information search is performed between two labeling sets.
FIG. 3 is a flowchart showing processing operations of a label score calculation unit and an information resource score calculation unit 32;
4 is a conceptual diagram for explaining processing (label score calculation processing) from step S2 to step S4 in FIG. 3;
5 is a conceptual diagram for explaining processing (information resource score calculation processing) from step S5 to step S7 in FIG. 3;
FIG. 6 conceptually shows a case where information resources are searched from a plurality (for example, four) of labeling sets A to D, and integrates information search results obtained from the labeling sets of individual search targets. The figure for demonstrating a case.
FIG. 7 is a diagram conceptually showing a case where information is searched and collected using the information search method of the present invention.
FIG. 8 is a diagram conceptually illustrating a case where information search is performed via a network, for example, using an information search apparatus according to a second embodiment of the present invention.
FIG. 9 is a diagram schematically showing a configuration of a main part of an information search device according to a second embodiment of the present invention.
FIG. 10 shows the operation of the information search apparatus of FIG. 9 in a user interface screen presented in the user interface unit (a pallet for displaying a list of information resources related to a label and a palette for manipulating labels and collectors). The figure which showed notionally based on the relationship with the collector mainly used for information retrieval.
FIG. 11 is a diagram showing an example of a pallet screen presented in the user interface unit.
FIG. 12 is a diagram showing an example of a finder presentation screen presented in the user interface unit when creating a collector, showing a case where a search result (candidate list) is not presented.
FIG. 13 is a diagram showing an example of a finder presentation screen presented in the user interface unit when creating a collector, showing a case where a search result (candidate list) is presented.
FIG. 14 is a diagram for explaining a collector used for information retrieval;
FIG. 15 is a display example when the contents of the collector are displayed in the finder.
FIG. 16 is a diagram for specifically explaining score calculation of candidate information resources, taking as an example a case where information search is performed on two other labeling sets B and C using a collector created from the labeling set A; Figure.
FIG. 17 is a diagram schematically illustrating a configuration of a labeling server.
FIG. 18 is a diagram showing a screen display example of a finder having an importance setting area for setting the importance of each information resource.
[Explanation of symbols]
1 ... Information search processing section
10 ... Information resource list acquisition unit for search
20 ... Labeling set selection unit 20
30 ... Information search section
31 ... Label score calculation unit
32. Information resource score calculation unit
33 ... Candidate information resource list creation unit
40 ... Candidate information resource presentation section
411 ... Labeling server
412 ... Labeling database
501 ... User interface section
502... Processing unit
503 ... Labeling processing section
504 ... Collector creation processing unit
505 ... Information search processing unit
506: Data storage unit
507 ... Transmitter / receiver
510 ... Label definition part
520 ... Label giving unit
530 ... Collector search formula
540 ... Main list acquisition unit
550 ... Main list operation unit
560 ... Collector selector
570 ... Information search section
580 ... Candidate information resource list presentation unit
610 ... pallet
620 ... Finder
630, 710 ... collector

Claims (13)

Storage means for storing a plurality of labeling sets composed of a plurality of information resources characterized by the correspondence between the information resources and the labels assigned to the information resources ;
Information retrieval processing means;
An information search method in an information search device comprising:
The information search processing means is based on a logical expression of a label specified by using a label assigned to an information resource in a labeling set of one search source among the plurality of labeling sets. Obtaining a search information resource group including one or more information resources characterized by a logical expression of the label from a labeling set ;
The information search processing unit selects one or a plurality of labeling sets to be searched from among the labeling sets other than the search source labeling set among the plurality of labeling sets stored in the storage unit. A labeling set selection step;
The information retrieval processing means, each labeling set of the search target, for each label in the labeling set of the search objects are assigned to the information resources of the search information resource group, and the search information resource group A label evaluation step for obtaining a first evaluation value for evaluating a connection with the label;
The search that is attached to the information resource of the search information resource group among the information resources other than the search information resource group in the labeling set for each labeling set of the search target by the information search processing unit For an information resource with a label in the target labeling set, a second evaluation value for evaluating the degree of association with the search information resource group is obtained using the first evaluation value of the label. The required information resource assessment step;
An information resource selection step in which the information search processing means selects an information resource related to the search information resource group from the search target labeling set based on the second evaluation value ;
The information search processing means presenting the information resource selected in the information resource selection step ;
Information retrieval method including Said information resource selection step, the second evaluation value of the information resource determined from the respective labeling group of the plurality of search target, the weight coefficient predetermined for each of the labeling group of the plurality of search target The information search method according to claim 1, wherein information resources related to the information resource group for search are selected based on the integration result. The label evaluation step, the label attached to the information resources of the search information resource group, by accumulating the predetermined value per one information resource 1 attached was the search for information resources groups that label, the 2. The information search method according to claim 1, wherein a first evaluation value of the label is calculated. The label evaluation step calculates the first evaluation value for a label in the search target labeling set based on the importance of the information resource of the search information resource group to which the label is attached. The information search method according to claim 1. Said information resource evaluation step, information resources other than the search information resource group in the labeling set of the search target, by accumulating the first evaluation value of the label that is attached to the information resource, information search method according to claim 1, wherein the calculating the second evaluation values of the information resource. Said information resource evaluation step, information resources other than the search information resource group in the labeling set of the search target, and severity of the first evaluation value and the label of the label that is attached to the information resource the basis of the information retrieval method according to claim 1, wherein the calculating the second evaluation values of the information resource. Storage means for storing a plurality of labeling sets composed of a plurality of information resources characterized by the correspondence between the information resources and the labels assigned to the information resources ;
Based on the logical expression of the label specified using the label given to the information resource in the labeling set of one search source among the plurality of labeling sets, the label of the label is extracted from the labeling set of the search source . An acquisition means for acquiring an information resource group for search including one or more information resources characterized by a logical expression ;
Labeling set selection means for selecting one or more labeling sets to be searched from among the labeling sets other than the search source labeling set among the plurality of labeling sets stored in the storage means;
For each label set to be searched, for each label in the search target label set attached to the information resource of the search information resource group , evaluation of the connection between the search information resource group and the label Label evaluation means for obtaining a first evaluation value to be performed;
For each labeling set to be searched, among the information resources other than the information resource group for search in the labeling set, a label in the labeling set of the search target attached to the information resource of the information resource group for search Information resource evaluation means for obtaining a second evaluation value for evaluating the degree of association with the information resource group for search using the first evaluation value of the label for the information resource to which
Information resource selection means for selecting an information resource related to the search information resource group from the labeling set to be searched based on the second evaluation value;
Presenting means for presenting the information resource selected by the information resource selecting means ;
An information retrieval apparatus comprising: The information resource selection means, said second evaluation value of the information resource determined from the respective labeling group of the plurality of search target, the weight coefficient predetermined for each of the labeling group of the plurality of search target 8. The information search apparatus according to claim 7, wherein the information resources associated with the search information resource group are selected based on the integration result based on the integration result. The label evaluation means, for labels applied to information resources of the search information resource group, by accumulating the predetermined value per one information resource 1 attached was the search for information resources groups that label, the The information search apparatus according to claim 7, wherein a first evaluation value of the label is calculated. The label evaluation unit calculates the first evaluation value for a label in the search target labeling set based on the importance of the information resource of the search information resource group to which the label is attached. The information search device according to claim 7. The information resource evaluation unit, the information resources other than the search information resource group in the labeling set of the search target, by accumulating the first evaluation value of the label that is attached to the information resource, information retrieval apparatus according to claim 7, wherein the calculating the second evaluation values of the information resource. The information resource evaluation unit, the information resources other than the search information resource group in the labeling set of the search target, and severity of the first evaluation value and the label of the label that is attached to the information resource the basis of the information retrieval apparatus according to claim 7, wherein the calculating the second evaluation values of the information resource. Computer
Storage means for storing a plurality of labeling sets composed of a plurality of information resources characterized by the correspondence between the information resources and the labels assigned to the information resources ;
Based on the logical expression of the label specified using the label given to the information resource in one labeling set of the search source of the plurality of labeling sets, the label of the label is extracted from the search source labeling set. An acquisition means for acquiring a search information resource group including one or more information resources characterized by a logical expression ;
Labeling set selection means for selecting one or a plurality of labeling sets to be searched from among the labeling sets other than the search source labeling set among the plurality of labeling sets stored in the storage means;
For each label set to be searched, for each label in the search target label set attached to the information resource of the search information resource group , evaluation of the connection between the search information resource group and the label Label evaluation means for obtaining a first evaluation value
For each labeling set to be searched, among the information resources other than the information resource group for search in the labeling set, a label in the labeling set of the search target attached to the information resource of the information resource group for search Information resource evaluation means for obtaining a second evaluation value for evaluating the degree of association with the information resource group for search using the first evaluation value of the label for the information resource to which
Information resource selection means for selecting an information resource related to the search information resource group from the labeling set to be searched based on the second evaluation value;
Presenting means for presenting the information resource selected by the information resource selecting means;
A machine-readable recording medium in which a program for functioning as a computer is recorded.

Images