Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1998-06-26
2002-05-14
Choules, Jack (Department: 2177)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C707S793000, C707S793000, C707S793000, C342S357490, C342S357490, C342S352000, C709S203000
Reexamination Certificate
active
06389409
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to daft retrieval and learning systems.
BACKGROUND OF THE INVENTION
In the context of distributed information systems (e.g the Internet), there is a need to provide end users with a centralized access and search services to information residing in multiple heterogeneous on-line catalogs. These on-line catalogs should be viewed by the users as if they were using the very same access method, information classification and nomenclature. This concept is called ‘information integration’ and is the subject of several research and development effort. Among them are:
Stanford University, Knowledge Systems Laboratory (KSL) Ontology Server Projects
Microelectronics and Computer Technology Corporation (MCC)—Infoslouth
There are three main problems associated with information integration:
1) Different conceptualization systems—providing access to relevant information which is accessible through different classification methods and described using non identical nomenclatures. This means bridging the gap between the different conceptualization systems—the one used by the user to describe his query and those used by each of different information resources. These conceptualization differences range from classification method to nomenclature (e.g. the user is looking for “RS232 Cable for Printer” which is listed in one on-line catalog under the name “RS232 cable” in the sub section “Accessories” in the super section “Painters” and in another on-line-catalog under the name “Printer cable” in the section “Hardware accessories”). This is a very tough task, since it involves the formalization of “knowledge”.
2) Resource selection—deciding which of the available information resources is relevant for a specific information request (e.g. there is no point in accessing resources providing information about restaurants when the user looks for automobile). Indeed, in the domain level, it is an easy task However, in larger arrays of information resources from similar domains, the problem becomes harder.
All the research projects listed above deal with different aspects of these problems making different assumptions on the environment. However, to-date, there are no general purpose information integration systems at all. There are two main reasons for this:
1. There are no automatic mechanisms to “connect” to new information resources. Current solutions to the task of connecting to information resources are based on the assumption that “someone” (either the information requester or the information provider) provides information source “wrapper” that enables “smooth” integration to the data.
2. There is no way to automatically a create large scale conceptualization system. The current solution to the problem of creating a common unified conceptualization system is a manual solution provided by the Knowledge System Laboratory (KSL) at Stanford University. The KSL staff has developed a set of tools and services to support the process of manually building and achieving consensus on a common shared conceptualization system (termed “Ontology”.
It is only natural, then, that the lack of a real world conceptualization system adversely affects both the quality of the information being retrieved (recall and precision) and the quality of the user-computer interaction. That is, real world information integration requires the automatic acquisition of conceptual knowledge base (conceptulization system).
In recent years, the task of automatic knowledge acquisition was usually approached by corpus-based NLP. Free text documents were used as a source for learning different relations between words (e.g., contextual similarity).
SUMMARY OF THE INVENTION
The emergence of a global standard computer network, and more specifically, the Internet, has led to the proliferation of classified on-line catalogs. This enables to use the information navigation systems. One of the innovations of the present invention is the usage of the knowledge embedded in these very navigation systems as a new source for the knowledge acquisition task in order to generate a so called unified classification information graph. Information navigation systems, by their nature, imply hierarchy relations between categories, hence enable to learn more precise category-relations information then free text does. The categories and the hierarchy relations between categories is utilized in the process of generating the unified classification information graph.
The present invention offers how to overcome the difficulties in the usage of the multiple resources (e.g. the same piece of information may be expressed in word order or levels of abstraction) so as to generated the desired unified classification information graph.
Since on-line catalogs are by nature subject to frequent (and occasionally also major) changes (e.g. new products/categories are added and/or others are deleted) it is important to sure that all or at least most of the modifications that occurred in the online catalogs will be reflected in the resulting unified classification information graph. Accordingly, one of the important advantages of the system is the dynamic nature thereof, i.e. the ability to dynamically scan the multiple information resources and update, whenever required, the resulting unified information graph.
The invention fulfills thus a long felt need by providing a system and method for obtaining and integrating multiple classification information resources using a single unified access interface.
By one aspect, the invention provides for a method for dynamically obtaining a unified classification information graph which provides a navigation system for a user to access sought information, comprising:
providing a multiple information resources that include a respective hierarchy of categories each of which associated with a category; leaf categories in said hierarchy being connected to information pages;
generating a unified classification information graph utilizing at least the hierarchy of categories and the categories of said multiple information resources; said unified classification graph includes hierarchy of unified categories; leaf unified categories in said hierarchy being connected to information pages;
whereby, information pages accessible through the hierarchy of said multiple information resources are also accessible through the hierarchy of said unified classification information graph.
By one embodiment, said step (a) includes providing at least some of said multiple information resources that are located in sites of the Internet.
By another embodiment, said step (a) includes providing at least some of said multiple information resources that are located in databases.
By still another embodiment said step (a) includes providing at least some of said multiple information resources that are located in on-line catalog.
Still further there is provided the step of associating categories in said hierarchy of categories in said multiple information resources with hyper-links.
Yet still further there is provided the step of associating categories in said hierarchy of categories in said multiple information resources with menus.
By one embodiment said step (b) includes:
(i) initalization so as to generate respective link graph that correspond to each information resource; said link graph includes link graph categories;
(ii) normalizing the link graph categories so as to generate classification graph that includes classification graph categories; and
(iii) unifying said classification graph so as to generate said unified classification information graph.
By this embodiment there is further provided the step of providing a URL pointer of said on-line catalog for generating said ink graph.
By another aspect the invention provides for a machine having a memory which contains data representing a unified classification information graph which was generated by the above method.
Still further, there is provided a memory for storing data for access by an application program, which approgram is accessed by a user through a user inter
Horovitz Oren
Karov Yael
Agentics Ltd.
Choules Jack
Inventek
Lewis Cheryl
Rosenfeld Dov
LandOfFree
Method for obtaining unified information graph from multiple... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for obtaining unified information graph from multiple..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for obtaining unified information graph from multiple... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2900147