# US 2006/0041604 A1 · Combined classification based on examples, queries, and keywords

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2006/0041604 A1
- Document ID: US20060041604A1
- Status: Published application
- Kind: A1
- Application number: 10/923,553
- Filed: 2004-08-20
- Published: 2006-02-23
- Earliest claimed: 2004-08-20
- Assignees: Sap SE
- Inventors: Thomas Peh
- Web page: https://patentgrep.com/patent/US20060041604A1

## CPC

- G06F 16/353: into predefined classes
- G06F 16/38: Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- Y10S 707/99932: Access augmentation or optimizing
- Y10S 707/99935: Query augmenting and refining, e.g. inexact access
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y10S: TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS

## Abstract

A document classification module may include a classification module that enables a user to classify documents using a combination of example-based classification (EBC) and query-based classification (QBC) methods. The user may also enter keywords with positive, negative, or null weights to fine tune the results of the classification process.

## Claims

### Claim 1

A method comprising:
identifying terms in a document to be classified;
generating one or more class candidates for said document from a plurality of classes based on a query;
assigning a first rank to each of the class candidates based on at least one of a positive-weight, negative-weight, and null-weight keyword associated with said class candidates; and
assigning the document to one of the class candidates.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
assigning one or more training documents to one or more of said plurality of classes.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
assigning a second rank to each of the class candidates based on training documents associated with the remaining class candidates.

### Claim 4 (depends on 3)

The method of claim 3, wherein said ranking the class candidates based on the training documents comprises setting a vector space component corresponding to a null-weight keyword to zero.

### Claim 5 (depends on 3)

The method of claim 3, further comprising:
calculating a mean rank from the first rank and the second rank associated with each of the one or more class candidates; and
assigning the document to one of the class candidates based on the mean rank.

### Claim 6

An article comprising a machine-readable medium including machine-executable instructions to cause a machine to:
identify terms in a document to be classified;
generate one or more class candidates for said document from a plurality of classes based on a query;
assign a first rank to each of the class candidates based on at least one of a positive-weight, negative-weight, and null-weight keyword associated with said class candidates; and
assign the document to one of the class candidates.

### Claim 7 (depends on 6)

The article of claim 6, further comprising instructions to cause the machine to:
assign one or more training documents to one or more of said plurality of classes.

### Claim 8 (depends on 7)

The article of claim 7, further comprising instructions to cause the machine to:
assign a second rank to each of the class candidates based on training documents associated with the remaining class candidates.

### Claim 9 (depends on 8)

The article of claim 8, wherein said ranking the class candidates based on the training documents comprises setting a vector space component corresponding to a null-weight keyword to zero.

### Claim 10 (depends on 8)

The article of claim 8, further comprising instructions to cause the machine to:
calculate a mean rank from the first rank and the second rank associated with each of the one or more class candidates; and
assigning the document to one of the class candidates based on the mean rank.
