 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method and apparatus for displaying items of information, e.g., data or files, organized in a hierarchical structure, the method and apparatus being capable of displaying portions of the hierarchical structure with
which the user is concerned, displaying items of user's concern collectively, and displaying classified trees of items adapted to the user's intention.
2. Description of the Prior Art
Since the advent of compact capacious storage devices such as optical disk memories in recent years, the management and search of a vast amount of information have become routine tasks of the office job. For example, the storing, searching and
management activities for a large-scale database, which used to be implemented by specialists, are now carried out by means of an office-oriented document filing system that is designed to be run by the user.
In the document filing system, documents and information are treated in a hierarchical structure in terms of key words and/or concepts in many cases. The hierarchical structure is displayed in the form of menu, on which the user selects items
sequentially to reach the intended document or information. However, this menu-based search scheme is sometimes deficient in that if the user once steps in a wrong path, the user is obliged to retry the search from the beginning.
A conventional filing system that overcomes the above-mentioned problem by the application of the knowledge base is disclosed in U.S. Pat. No. 4,868,733 entitled "DOCUMENT FILING SYSTEM WITH KNOWLEDGE-BASE NETWORK OF CONCEPT INTERCONNECTED BY
GENERIC, SUBSUMPTION AND SUPERCLASS RELATIONS".
Another proposal for displaying and browsing through a tree of concepts, which are stored in the form of hierarchical structure in a knowledge base, is described in Japanese patent publication JP-A-1-140332.
The following explains in brief the manner of knowledge expression in the knowledge base pertinent to the above-mentioned prior art. FIG. 2 is a diagram showing the manner of knowledge expression for the knowledge base, in which each ellipse
represents a concept and each arrow represents the relation between two concepts. In the knowledge base, concepts are organized in a hierarchical structure based on the is-a relation and part-whole relation. All concepts in the knowledge base are
linked together in is-a relations below the superlative concept "UNIVERSAL" 201. The is-a relation represents a relation "concept X is (a) concepts Y". Seen from a concept, another concept located above it is called a "super-concept" and another
concept located below it is called a "sub-concept". A concept that shares a same super-concept with another concept is called an "appositive concept". For example, in the figure, "person" 212 has a super-concept "intellectual entity" 205, has
sub-concepts "female" 220 and "Tanaka Yoko" 213, and has an appositive concept "organization" 206.
Some concepts are linked with other concepts in a part-whole relation. The part-whole relation represents a relation "concept X is part of concept Y". In the part-whole relation, a super-concept, i.e., a concept located above another one is
called a "whole-concept" and a sub-concept, i.e., a concept located below another one is called a "part-concept". For example, "X-laboratory" 209 is a part-concept of "A-company" 208.
Each concept can have multiple is-a relations and multiple part-whole relations. For example, the "organization" 206 has two super-concepts of "intellectual entity" 205 and "place" 214 in its is-a relation.
In the knowledge base, things are expressed in terms of two concepts and one relation, i.e., binary relation. First, two generic concepts are related in terms of a generic relation. For example, "person" 212 and "nation" 215 are linked by a
generic relation "nationality" 218. Next, two specific concepts are linked in terms of a instance relation. In order for two concepts to establish an instance relation, their super-concepts must be linked by a generic relation of the same name. For
example, concepts "Tanaka Yoko" 213 and "Japan" 216 are linked by the instance relation "nationality" 219.
By tracing these relations that link concepts stored in the knowledge base, the user can "browse" through the concepts. It has become possible to make access to knowledge by entering a query, as described in the above-mentioned U.S. Pat. No.
4,868,733. For example, such queries as "article #0468 that covers the workstation developed by B-company seated in Tokyo" and "what company published the article that deals with the subject of information retrieval system ?" are entered to the
knowledge base to make access to detailed information.
There have been proposed several methods of displaying the "concept network" of knowledge base that is based on the above-mentioned expression, as described in Japanese patent publication JP-A-1-140332 for example. According to the proposed
display method, the system initially displays a hierarchical tree headed by the superlative concept "UNIVERSAL". The system includes a function of showing sub-concepts with a depth of m and a number of branches of n specified by the user for one concept
existing on the tree, and a function of hiding sub-concepts of one concept. Further included in the system is a function of listing sub-concepts immediately below one concept, allowing the user to select one with a pointing device and showing the
selected concept to the displayed hierarchical tree, a function of searching for concepts based on the matching of a partial character string, and a function of displaying super-concepts of one concept.
However, the foregoing prior art methods have the following problems. One problem is that because of a hierarchical tree displayed in one window, when the user intends to hide a sub-concept with the hiding function, the tree expands as the user
traces down the tree, making it difficult to grasp the whole tree and find the intended concept. Another problem is that in searching for a sub-concept by way of several intermediate concepts, the user is obliged to repeat the operations in such a
manner of displaying sub-concepts located immediately below one concept and selecting one, and displaying sub-concepts located immediately below the selected concept and selecting one, and so on. In this case, if the user takes a wrong path, the
hierarchical tree becomes intricate and the search operation has to be retried from the beginning. Still another problem is that although super-concepts seen from one concept up to the superlative concept "UNIVERSAL" can be listed, the user who intends
to view appositive concepts of one super-concept needs to select each super-concept and list sub-concepts located immediately below it. On this account, it is difficult for the user to grasp the taxonomic system and find the position of definition of
concepts.
It is possible for a knowledge base to preserve queries in the form of macros, as proposed in U.S. patent application Ser. No. 07/430,241 entitled "SYSTEM FOR DISPLAYING CONCEPT NETWORKS". However, since these macro-wise queries are stored in
a file different from that used by the knowledge base in display, the user is obliged to reference to the file and open the window again.
The foregoing prior art systems are merely capable of displaying hierarchical trees of hierarchical structures that are defined in terms of the is-a relation, disallowing the display in classified forms adapted to individual users. Although the
prior art systems can display the instance relation to one concept in the form of a frame or can list instance relations between one generic concept and its sub-concepts, these systems are merely capable of displaying hierarchical trees of hierarchical
structures that have been defined in advance in terms of the is-a relation. On this account, when the user intends to search for companies through the classification based on the seat, for example, the system can possibly offer only a display of
companies that are classified based on the industrial category.
Prior art methods for displaying an appropriate window on the display screen through the operation of nodes are as follows.
A publication "Macintosh user's guide for desktop Macintosh computers" published by Apple Japan, Inc. has the following description in the sections "How to use view menu" (p. 120) and "Displaying hierarchy in a folder" (p. 238): "By
single-clicking the triangular mark on the left of the folder icon, files or low-order folders below the folder are displayed. (Icons of files are different depending on each application that have created the files.) By double-clicking the folder icon
(or clicking it and selecting "open . . . " from the menu, another window is displayed and files or low-order folders below the folder are displayed". Another section "open outward folder" (p. 238) describes: "By selecting the title of window of a
folder, high-order folders above the folder are displayed."
A publication "Microsoft Windows Version 3.1 Getting Started with Microsoft Windows For Microsoft Windows Operating System Gateway 2000 Edition" has the following description in the section "File Manager" (pp. 37-43): "Directory names and
associated directory icons are displayed in the form of a hierarchical tree. By specifying a directory in the tree, files or sub-directories under that directory are listed. (Icons of files are different depending on each application that have created
the files.)"
In the foregoing prior art systems, when the user intends to display items that are always within the scope of one's concern, the user is obliged to trace menus from the superlative concept (node), carry out the character string matching
operation, and repeat the same operation in every time. Items of user's concern are scattered among unnecessary items, and the user is obliged to scroll the window or switch the window frequently. Reserved queries and glossaries are displayed in
different windows, and the user is obliged to open and close several windows. Items are displayed solely based on the hierarchical structures that are provided for the database (knowledge base), and the user cannot choose the taxonomic system that suits
the user's purpose.
SUMMARY OF THE INVENTION
The present invention is intended to solve the foregoing prior art problems, and its prime object is to provide a method and apparatus for displaying collectively items of information with which each user is concerned and displaying a
hierarchical tree of items through a single user's operation.
Another object of this invention is to provide an information management method and apparatus capable of treating queries, glossaries and images in a unified manner based on a single hierarchical structure.
Still another object of this invention is to provide an information display method and apparatus capable of revising a displayed hierarchical tree of information to meet the user's intention.
In order to achieve the above objectives, the method and apparatus of this invention, which is intended to enable the user to browse through displayed items organized in a hierarchical structure, are designed to collect a certain subset of items
(nodes) from a set of items on a hierarchical tree to form a partial hierarchical tree that is represented by a representative node and display such representative nodes in the form of a hierarchical tree.
This invention is applicable to any set of items that is treated in a hierarchical structure. Examples of application include a computer file or data system organized in a hierarchical structure, and a concept network consisting of concepts
having the is-a relation with each other.
According to this invention for displaying the hierarchical structure of numerous concepts (data) by use of a multi-window function for example, a partial hierarchical tree with which the user is concerned is treated as a set and it is listed or
displayed as a hierarchical tree in a different window so that only the fields of the user's concern are displayed collectively. Glossaries, which are used to deal with character strings of concepts, queries and images are treated, displayed and
manipulated based on the hierarchical structure. Attributes of each concepts (data) are saved so that concepts (data) can be displayed by being classified based on attributes selected by the user.
The inventive method and apparatus are designed to mark a node (concept and data) of user's concern on the hierarchical tree, save information of a partial hierarchical tree of the displayed node, and display the node in a separate window. At
the next event of information retrieval, the system lets the user select a node among the collected nodes of user's concern thereby to pop up a window for displaying the reserved partial hierarchical tree of that node. Since the collected nodes of
user's concern and information of partial hierarchical trees of these nodes are saved, the same hierarchical tree as of the preceding session can be displayed in the next occasion.
Each query or glossary is treated as a node in the same hierarchical structure, and each image is reserved as an additional item of a node. Queries, glossaries and images may be displayed as visual nodes so that they can be distinguished from
other nodes.
An attribute information pertinent to a concept (data) is reserved in terms of a generic relation that links two generic concepts and an instance relation that links two instance concepts corresponding to that generic relation. In consequence,
it is possible to classify concepts based on the instance relation corresponding to the generic relation selected by the user. A resulting classified tree is treated as a node in the same hierarchical structure together with the representative node of
user's concern.
This invention introduces the idea of the conceptual representative node in which a partial hierarchical tree of concepts (data) is treated as a set, and consequently it is possible to gather partial hierarchical trees of user's concern.
Consequently, the user can easily search for an intended concept by browsing through fields of user's concern.
The inventive method and apparatus are capable of classifying concepts (data) not only based on the is-a relation defined in advance, but also based on a relation (attribute) appended to them, allowing the user to alter the hierarchical tree to
meet the user's intention.
The inventive method and apparatus allow the user to add user-oriented trees and queries to a hierarchical tree of concepts (data), and consequently the user can treat, search and display these items in a unified manner.
BRIEF DESCRIPTION
OF THE DRAWINGS
FIG. 1 is a diagram showing an example of the display screen for the interactive operation of the information system based on an embodiment of this invention;
FIG. 2 is a diagram showing an example of the knowledge expression based on this invention;
FIG. 3 is a diagram showing the software module organization of the information system based on this embodiment;
FIG. 4 is a diagram showing the system configuration of this embodiment;
FIG. 5 is a diagram showing a record of information of concepts;
FIG. 6 is a diagram showing a record of the hierarchicay structure of concepts;
FIG. 7 is a diagram showing a record of information of relations (attributes);
FIG. 8 is a diagram showing a record of generic relations and instance relations;
FIG. 9 is a diagram showing a record of concepts and images in pairs;
FIG. 10 is a diagram showing a record of concepts and glossaries in pairs;
FIG. 11 is a diagram showing the initial menu;
FIG. 12 is a diagram showing a display of the sub-conceptual window including an associated menu;
FIG. 13 is a diagram showing an example of display when the sub-concepts window is open;
FIG. 14 is a diagram showing an example of display of the sub-concepts window when the user browses through lower sub-concepts;
FIG. 15 is a diagram showing an example of display of the sub-concepts window when the user browses through much lower sub-concepts;
FIG. 16 is a diagram showing an example of display after the user has revised the display of sub-conceptual window from the sub-concepts window;
FIG. 17 is a diagram showing an example of display after the user has revised the display of sub-conceptual window at the definition of concepts as conceptual representative nodes;
FIG. 18 is a diagram showing an example of display after the user has revised the display of conceptual window at the definition of concepts as conceptual representative node;
FIG. 19 is a diagram showing an example of display after the user has selected a conceptual representable node;
FIG. 20 is a diagram showing an example of display of the super-concepts window;
FIG. 21 is a diagram showing an example of display of the query edit window including an associated menu;
FIG. 22 is a diagram showing an example of display of the query edit window;
FIG. 23 is a diagram showing the window for entering the name of a query representative node;
FIG. 24 is a diagram showing an example of display of the classification window including an associated menu;
FIG. 25 is a diagram showing an example of display of the axis setting window;
FIG. 26 is a diagram showing an example of entry to the axis setting window;
FIG. 27 is a diagram showing an example of display of the classification window;
FIG. 28 is a diagram showing an example of display off the axis setting window;
FIG. 29 is a diagram showing an example of entry to the axis setting window;
FIG. 30 is a diagram showing an example of display of the classification window;
FIG. 31 is a diagram showing an example of entry to the axis setting window;
FIG 32 is a diagram showing an example of display of the classification window;
FIG. 33 is a diagram showing an example of entry to the attribute selection window;
FIG. 34 is a diagram showing an example of display of the classification window;
FIG. 35 is a diagram showing the window for entering the name of a classification representative node;
FIG. 36 is a diagram showing an example of display of the sub-conceptual window after a classification representative node has been defined;
FIG. 37 is a diagram showing an example of display of the conceptual window after a classification representative node has been defined;
FIG. 38 is a flowchart showing the operation when the user browses through a hierarchical tree;
FIG. 39 is a flowchart showing the operation when the user browses through the sub-conceptual window;
FIG. 40 is a flowchart showing the operation when the user browses through the sub-concepts window;
FIG. 41 is a flowchart showing the conceptual node define delete operation;
FIG. 42 is a flowchart showing the operation when the user browses through the super-concepts window;
FIG. 43 is a flowchart showing the query edit operation in the query edit window;
FIG. 44 is a flowchart showing the display operation for the classification window; and
FIG. 45 is a flowchart showing the operation for the axis setting window.
DESCRIPTION OF THE PREFERRED EMBODIMENT
An embodiment of this invention will be described with reference to the drawings.
FIG. 4 shows the system configuration of a filing system to which the information display method and apparatus based on an embodiment of this invention is applied. The system includes a CPU 401, a main memory 405, a group of magnetic disk units
406, a display unit 402, a keyboard 403, and a pointing device 404. The system is designed to enter semantic information to a knowledge base and retrieve information from it for display. The knowledge base stored in the magnetic disk units 406 is
loaded into the main memory 405. The user makes access to the knowledge base through the operation of the display unit 402, keyboard 403 and pointing device 404, and the system displays requested knowledge in a form specified by the user on the display
unit 402. The inventive method and apparatus are used in the system in displaying retrieved knowledge.
FIG. 3 shows part of the software module organization of the knowledge base system that uses the concept network based on this embodiment. The system performs the information retrieval and display in response to the user's request and also the
storing and accumulation of fragmentary information, which is entered by the user, in the knowledge base 318 that is formed of the concept network.
A concept network memory 318 stores concept network data that is expressed in compliance with the knowledge expression model shown in FIG. 2. Stored data can be expressed in the text format that is compatible with external systems. Stored data
is loaded into the memory of a concept network manager 316 at the start-up of the system by being converted to have an internal data structure that enables high-speed browsing. The concept network data of the internal data structure is accessed by a
concept network editor 312, view generator 313 and concept retriever 314, and is rendered in the process so that it is useful for the edit, display and search of knowledge (concept network).
A glossary memory 317 stores character strings of concepts that are divided into several files on disks. Stored data is loaded into the memory of the concept network manager 316 in response to a request from the view generator 313. A document
image memory 319 stores images of documents.
An input/output control part 304, which consists of a window controller 306 and input analyzer 305, controls the interactive operation with the user. The input analyzer 305 analyzes input information from the keyboard and mouse device through
the process of an interactive display control part 307, which consists of a concept network edit controller 308, a concept network viewer 309, a query editor 310 and an image viewer 311.
FIG. 5 through FIG. 10 show the internal data structure of the concept network data in the files of this system. Shown in FIG. 5 through FIG. 8 resemble the internal data structure described in the above-mentioned U.S. Pat. No. 4,868,733.
Specific data shown in FIG. 5 through FIG. 10 (not all data are shown) correspond to the data that represent the concept network shown in FIG. 2.
Shown in FIG. 5 is a correspondence table between node IDs (concept IDs) and node names (concept names). The table includes a C# field for the node ID, a CNAME field for the node name, and a P/S field for the flag indicative of whether or not
the node name is to be displayed. The P/S flag is used to specify a node name to be displayed when same concepts have more than one name. This table differs from the counterpart of the U.S. Pat. No. 4,868,733 in the addition of a C/Q/T field. In
this field, a C flag indicates that the node ID is a concept ID, a Q flag indicates that the node ID is a query ID, and a T flag indicates that the node ID is a classification ID. For example, the node ID "1" has a concept name of "UNIVERSAL", and the
node ID "5" has two node names "person" and "human" and a P/S flag "P" indicative of the concept name "person" to be displayed. The node ID "4" has a C/Q/T flag "Q" indicative of a query ID having a name of "article of information retrieval system", and
the node ID "8" has a C/Q/T flag "T" indicative of a classification ID having a name of "company classified by seat".
FIG. 6 is a table indicating the rank of concepts in the is-a relation. The table includes a PARENT field for entering a concept ID that indicates a super-concept, i.e., parent concept, in the is-a relation, and a CHILD field for entering a
concept ID that indicates a sub-concept, i.e., child concept, in the is-a relation. For example, the concept "organization" of concept ID "3" has a parent concept that is concept "intellectual entity" of concept ID "13".
FIG. 7 is a table containing names of relations that appear in the concept network. The table includes a RS# field that indicates the ID of relation name, a RSNAME field that contains a relation name, a LR field that contains a character string
indicative of the relation between two concepts seen from one concept as the subject, and a RL field that contains a character string indicative of the relation seen from the other concept. For example, RS# ID "7" has a relation name "nationality" and
two character strings "is nationality of" and "nationality is" indicative of the relation.
FIG. 8 is a table for accumulating generic relations and instance relations that have been defined. The table includes a R# field for entering a relation ID, a RS# field for entering a relation name ID defined in the table of FIG. 7, a CL field
for entering a concept ID of the subject concept when the character string defined in the LR field of the table of FIG. 7 is used, a CR field for entering a concept ID of the subject concept when the character string defined in the RL field of the table
of FIG. 7 is used, and a G/I field for entering a flag G indicative of a generic relation or a flag I indicative of an instance relation. For example, relation ID "12" is for the relation named by relation name ID "7" and is a generic relation (flag G)
between concept "5" and concept "6". In the table of FIG. 7, the two character strings in the LR and RL fields at the relation name ID "7", i.e., "nationality is" and "is nationality of", are obtained. As another example, relation ID "18" is for the
relation named by relation name ID "7" and is an instance relation (flag I) between concept "10" and concept "7". Specifically this relation ID defines the instance relation accompanied by two statements "the nationality of Tanaka Yoko is Japan" and
"Tanaka Yoko has a nationality of Japan". Part-whole relations are treated as special cases of generic relations in the tables of FIG. 7 and FIG. 8.
FIG. 9 is a table for containing concept IDs and image IDs, indicating whether or not an image of each concept is defined.
FIG. 10 is a table for containing concept IDs and glossary IDs, indicating whether or not a glossary of each concept is defined. Glossaries are stored in the form of the S-expression of LISP. For example, a concept ID "3" for "organization"
defines a glossary of "4". Glossaries are defined as character strings that can be sub-concepts of the "organization" in the form of LISP-based expression such as:
(G#4 ("XX group", "ABC electric company", "XYZ bank"))
It is also possible to define a character string among glossaries as a concept in the tables of FIG. 5 and FIG. 6. In this case, the concept ID is attached to it automatically. For example, when a character string "XYZ bank" is specified, it is
defined as a concept in the table of FIG. 5, with a new concept ID being affixed to it. This glossary is also defined in the table of FIG. 6 as a sub-concept of the concept which the glossary belongs, i.e., "organization" at C#="3", as a default option. The user can alter this parent concept "organization" to other concept such as "bank".
Queries having individual IDs are stored in the form of S-expression of LISP. For example, a query representative node "company that published information retrieval system" is described by the S-expression in terms of IDs of:
("company that published a information retrieval system" ("company" ("published document" ("article of information retrieval system"))
for query expression Q#4:
______________________________________ Query expression: company "article of information retrieval system" that is a published document of the company as follows. S-expression: (Q#4 (C#2 (R#17 (Q#2)) ______________________________________
The above query expression Q#4 includes a query expression Q#2 that represents "article of information retrieval system", and the query expression Q#2 is:
______________________________________ Query expression: article information retrieval system that is the subject of the article ______________________________________
This is described in terms of IDs of:
______________________________________ ("article of information retrieval system" ("article" ("subject" ("information retrieval system")) ______________________________________
as follows:
______________________________________ S-expression: (Q#2 (C#12 (R#23 (C#9)) ______________________________________
Classified trees having individual IDs are stored in the form of S-expression of LISP. For example, a classification tree "company classified by seat and industry" with the intention of classifying companies by seat, i.e., Japan, U.S.A., Britain
and other, and by industry, i.e., manufacturer, bank, service firm and other, is described in terms of IDs of:
______________________________________ ("company classified by seat and industry" (company (seat (concept-flag Japan U.S.A. Britain)) (industry (concept-flag manufac- turer bank service-firm))) ______________________________________
as follows:
______________________________________ (C#53 (C#2 (R#28 (C C#7 C#42 C#43)) (R#33 (C C#66 C#67 C#68)))) ______________________________________
A classified tree "employee of company classified by seat" is displayed as a combination of companies classified by seat and employees who are employed by individual companies. This classified tree is described in terms of IDs of:
______________________________________ ("employee of company classified by seat" (company (seat (concept-flag Japan U.S.A. Britain)) (industry concept-flag manufacturer bank service-firm))) (employee)) ______________________________________
as follows:
______________________________________ (C#54 (C#2 (R#28 (C C#7 C#42 C#43)) (R#33 (C C#66 C#67 C#68))) (R#55)) ______________________________________
The concept flag C indicates that the classification is based on the concept.
A classified tree "capital" with the intention of classifying capitals in the range 0-100 M-yen in a step of 50 M-yen is described in terms of IDs of:
______________________________________ ("employee of company classified by seat" (company (capital (numeric-flag 0 100 50)))) ______________________________________
as follows:
______________________________________ (C#55 (C#2 (R#29 (S 0 100 50)))) ______________________________________
The numeric flag S indicates that the classification is based on the numeric value.
As described above, queries and classified trees defined by the user are stored in the form of S-expression of LISP and displayed in the form of a tree through the analysis of the S-expression. For example, classified trees are accumulated only
for their classification manners in terms of the S-expression as explained above, and a classified tree is retrieved by another search in response to the user's request of tree-wise display.
Next, the data structure for display will be explained. In addition to the internal data of the system described above, tree structures that have been used by individual users are stored in the form of S-expression as follows.
______________________________________ (user ID (1 "UNIVERSAL" (1 "person" (1 "organization" (1 "university") (1 "document" (1 "patent" . . .) (0 "monograph" (0 "domestic paper" . . .) (0 "foreign "paper" . . .)) (0 "article | | |
