Method for detecting and extracting text data using database schemas

• US 5,717,913 A
• Filed: 01/03/1995
• Issued: 02/10/1998
• Est. Priority Date: 01/03/1995
• Status: Expired due to Term

• Current Assignee: University of Central Florida Research Foundation Inc. (State University System of Florida)
• Original Assignee: University of Central Florida (State University System of Florida)
• Inventors: Driscoll, James R.
• Primary Examiner(s): AMSBURY, WAYNE P
• Application Number:
  US08/368,045
• Time in Patent Office: 1,134 Days
• Field of Search: 395/600, 395/605, 395/606
• US Class Current: 1/1
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

An Information Filtering (IF) system for retrieving relevant text data from a data base document collection is disclosed. A user can use this system to access a dynamic data stream to retrieve relevant data such as accessing e-mail or a wire-service. Alternatively, a user can use the IF system to access an data storage archive such as electronically stored patents, journals and the like. The invention includes several steps. The first step has a user reduce the information they are interested in into a tangible form such as manually writing a natural language user need statement, or alternatively imputing the statement electronically into a computer file for storage. The next step is to create a filter window having an adjustable document viewing text length, that will be used to electronically scan through the database collection of documents in order to determine a relevancy value for each scanned document. The filter can be created several ways using synonym and domain lists. Alternatively, the synonym and lists for each document can be determined by Entity-Relationship (ER) modelling to generate a search schema. After documents receive relevancy values, the user is free to view only those documents having relevancy values that exceed a preselected threshold value. Documents can be ranked from most relevant to least relevant. Feedback information from viewing the retrieved documents can be used to update the synonym/domain lists of the filtering window to enhance the relevance retrieval of subsequent documents.

Claims

1. An Information Filtering (IF) system for retrieving relevant text from a database collection of documents comprising the steps of:

(a) defining an information interest as a natural language statement;

(b) creating a synonym list from each substantive word in the natural language statement;

(c) creating a domain list from the natural language statement;

(d) combining the synonym lists and the domain lists into a filter window;

(e) selecting a minimum threshold value for the filter window;

(f) scanning a first document having a first total length of a database collection with the filter window in order to calculate both a first value and a second value, wherein the first value is the number of matches between words in the synonym lists and corresponding words in the first document, and the second value is the number of matches between words in the domain lists and corresponding words in the first document;

(g) adding the first value to the second value to form a sum value, and dividing the sum value by the total length value of the first document to form a relevancy value for the first document; and

(h) repeating steps (a) through (g) for subsequent documents from the database collection if the relevancy value of each subsequent document is less than the minimum threshold value.

2. The Information Filtering (IF) system of claim 1, wherein the information interest of step

(a) is chosen from at least one of:

a topic of interest, a natural language query, and an area of interest.

3. The Information Filtering (IF) system of claim 1, wherein the database collection of step (f) includes:

a stream of data from a non-archive collection.

4. The Information Filtering (IF) system of claim 3, wherein the stream of data is selected from at least one of:

a wire service and e-mail.

5. The Information Filtering (IF) system of claim 1, wherein the database collection of step (f) includes:

an archive collection of data.

6. The Information Filtering (IF) system of claim 1, further including the steps of:

(i) modifying the synonym lists, the domain lists and the minimum threshold value for subsequent documents from the database collection if the relevancy value of each subsequent document is greater than the minimum threshold value, and repeating steps (a) through (g) for the subsequent documents.

7. The Information Filtering (IF) system of claim 1, wherein the substantive words further includes:

words that were not preselected for a stop-word list.

8. The Information Filtering (IF) system of claim 1, wherein the domain list includes:

words that answer the information interest.

9. An Information Filtering (IF) system for retrieving relevant data from a database comprising:

a database of documents;

a natural language need statement created from information interest of a user of the database;

a filter window of search words formed from the natural language need statement, wherein the filter window has a list of synonyms for each substantive word in the natural language need statement, and a list of domain words for satisfying portions of the natural language need statement;

means for scanning the database with the filter window to calculate relevancy values for each document in the database, wherein the relevancy values of each document includes;

a first value equal to the number of matches between words in the synonym lists and corresponding words in the first document, and a second value equal to the number of matches between words in the domain lists and corresponding words in the first document;

a sum value of the first value added together with the second value; and

a total length value for the each document, wherein the relevancy value is equal to the sum value divided by the total length value; and

means for ranking the relevancy values into a hierarchy list of documents from most important documents to the least important documents.

10. The Information Filtering (IF) system of claim 9, wherein the filter window further includes:

a synonym list and a domain list formed from an entity-relationship model.

11. An automatic method for retrieving relevant text information from a database using an Information Filtering (IF) system, comprising the steps of:

(a) generating a search schema filter by using an entity relationship model to generate the search schema, the model further including;

(i) generating a first entity;

(ii) generating a second entity; and

(iii) generating a relationship between the first entity and the second entity, and wherein the first entity, the second entity and the relationship, each include a single word; and

(iv) generating a synonym list and a domain list for each single word;

(b) searching database documents with the search schema filter for relevant documents; and

(c) ranking the relevant documents.