# US 2002/0055940 A1 · Method and system for selecting documents by measuring document quality

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

- Number: US 2002/0055940 A1
- Document ID: US20020055940A1
- Status: Published application
- Kind: A1
- Application number: 10/004,514
- Filed: 2001-11-02
- Published: 2002-05-09
- Earliest claimed: 2000-11-07
- Assignees: The Regents of the University of California
- Inventors: Charles Elkan
- Web page: https://patentgrep.com/patent/US20020055940A1

## CPC

- G06F 16/9536: Search customisation based on social or collaborative filtering
- G06F 16/9535: Search customisation based on user profiles and personalisation
- Y10S 707/99933: Query processing, i.e. searching
- Y10S 707/99943: Generating database or data structure, e.g. via user interface
- 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

The present invention relates to a system and method for classifying documents in order to select the most desirable documents of a group. Because quality is very difficult to distinguish by anyone other than a human being, this invention provides a system and method that will create a profile of what constitutes quality, then utilize this profile to allow a user to retrieve information that is desirable. A client is provided with items of data selected according to estimates computed using a profile of certain high-level criteria such as quality, interestingness, appropriateness, timeliness, humor, style of language, obscenity, sentiment, and any combinations thereof. These estimates are computed from low-level criteria such as length, vocabulary, fraction of words spelled correctly, title, author, reading grade level, average length of sentences, average length of words, usage of punctuation, usage of grammar, formatting, capitalization, source, display tags and any combinations thereof. The profile is learned automatically from labeled training examples. This system also relates to a method of obtaining and automatically associating a value to an item of data by obtaining items, obtaining labels for some items, selecting items of data with certain labels to form training sets, learning a profile using the training sets, and associating a value to another item of data using said profile. As such, the program is capable of learning to measure which items are of high quality and is capable of delivering only those items of data which would be of interest to a client.

## Claims

### Claim 1

A system for providing a client with at least one item of data according to said item of data's value, said system comprising;
a downloading component for obtaining at least one item of data from a source;
a classifier component for associating a value to each said item of data using a profile;
a training component that selects at least one of said item of data according to certain labels, said selected items of data being grouped to form training data;
a learning component that accepts said training data and automatically creates said profile; and
a presenter component for accepting a request from a client and transmitting said items of data selected according to said values.

### Claim 2 (depends on 1)

A system as in claim 1, further comprising a database for storing said items of data, said values and said profiles.

### Claim 3 (depends on 1)

A system as in claim 1, wherein said downloading component is effectuated at time intervals.

### Claim 4

A method of obtaining and automatically associating a value to an item of data comprising the steps of:
obtaining at least one item of data from a source;
obtaining labels for at least one of said items of data;
selecting items of data with certain labels to form training data;
creating a profile from said training data; and
associating a value to at least one of said items of data using said profile.

### Claim 5 (depends on 4)

The method of claim 4, further comprising the steps of:
receiving requests from clients; and
transmitting at least one item of data according to said requests and said associated values to said client.

### Claim 6 (depends on 4)

The method of claim 4, further comprising the steps of:
introducing at least one new item of data to said training data; and
generating a new profile from said training data.

### Claim 7 (depends on 4)

The method of claim 4, wherein said profile is automatically generated.

### Claim 8 (depends on 4)

The method of claim 4, wherein said value is an estimate of said label, for items without said label.

### Claim 9 (depends on 4)

The method of claim 4, wherein said profile specifies said associated value based on low-level features of said item selected from the group consisting of length, vocabulary, fraction of words spelled correctly, title, author, reading grade level, average length of sentences, average length of words, usage of punctuation, usage of grammar, formatting, capitalization, source, display tags and any combinations thereof.

### Claim 10 (depends on 4)

The method of claim 4, wherein said value is measured on a quantitative scale of measurement.

### Claim 11 (depends on 4)

The method of claim 4, wherein said value is measured on a categorical scale of measurement.

### Claim 12 (depends on 4)

The method of claim 4, further comprising storing said items of data in a database.

### Claim 13 (depends on 4)

The method of claim 4, further comprising storing said associated values in a database.

### Claim 14 (depends on 4)

The method of claim 4, wherein the step of obtaining labels for at least one of said items of data is accomplished by a human providing said labels.

### Claim 15 (depends on 4)

The method of claim 4, wherein said labels are selected according to levels of quality, interestingness, appropriateness, timeliness, humor, style of language, obscenity, sentiment, and any combinations thereof.

### Claim 16

A method for providing a client with at least one item of data according to said item of data's value, said method further comprising the computer implemented steps of:
obtaining labels for at least one of said items of data;
selecting items of data with certain labels to form training data;
creating a profile from said training data;
associating a value with at least one of said items of data using said profile;
accepting a request including value selection criteria from a client;
selecting at least one item of data according to said value selection criteria; and
transmitting said items of data to said client.

### Claim 17 (depends on 16)

A method as in claim 16, further comprising the steps of:
accepting a request including a value threshold from a client; and
retrieving items of data relative to said value threshold.

### Claim 18 (depends on 16)

A method as in claim 16, further comprising the steps of:
introducing at least one new item of data to said training data; and
generating a new profile from said training data.

### Claim 19 (depends on 16)

The method of claim 16, further comprising the step of associating a value with an item of data by retrieving a value previously associated with said item of data.

### Claim 20 (depends on 16)

The method of claim 16, further comprising the step of associating a value with an item of data by generating said value for said item of data using said profile.

### Claim 21 (depends on 16)

The method of claim 16, wherein said value is quantitative.

### Claim 22 (depends on 16)

The method of claim 16, wherein said value is categorical.

### Claim 23 (depends on 16)

The method of claim 16, further comprising storing said items of data in a database.

### Claim 24 (depends on 16)

The method of claim 16, further comprising storing said associated values in a database.

### Claim 25

A method for providing a client with at least one item of data according to said item of data's value, said method further comprising the computer implemented steps of:
obtaining labels for at least one item of data, wherein said item of data is information contained within an electronic bulletin board, and said labels designate level of quality, interestingness, appropriateness, timeliness, humor, style of language, obscenity, sentiment, or any combination thereof;
selecting said items of data with certain labels to form training data;
creating a profile from said training data;
associating a value to at least one of said items of data using said profile;
accepting a request including value selection criteria from a client;
selecting at least one item of data according to said values and said value selection criteria; and
transmitting said selected items of data to said client.

### Claim 26 (depends on 25)

The method of claim 25, further comprising storing said items of data in a database.

### Claim 27 (depends on 25)

The method of claim 25, further comprising storing said associated values in a database.

### Claim 28

A system for automatically retrieving and presenting a client with items according to their qualitative nature, comprising:
at least one computer system having at least one item of data available;
at least one access device for enabling said client to communicate with said computer system;
a classifier means effective to automatically associate a value to said items of data, wherein said value is indicative of the qualitative nature of said data;
a means for a client to provide a request for at least one said item of data according to criteria; and
a transmitting means adapted to present at least one said item of data to said client selected according to said criteria.

### Claim 29 (depends on 28)

A system as in claim 28, wherein said item of data is a text message.

### Claim 30 (depends on 28)

A system as in claim 28, wherein said value is selected from the group consisting of quality, interestingness, appropriateness, timeliness, humor, style of language, obscenity, sentiment, and any combinations thereof.
