System for processing sentiment-bearing text

• US 20060200342A1
• Filed: 04/14/2005
• Published: 09/07/2006
• Est. Priority Date: 03/01/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Ringger, Eric K., Gamon, Michael, Corston-Oliver, Simon H., Aue, Anthony
• Granted Patent: US 7,788,087 B2
• Time in Patent Office: Days
• US Class Current: 704/10
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique

Abstract

The present invention provides a system for identifying, extracting, clustering and analyzing sentiment-bearing text. In one embodiment, the invention implements a pipeline capable of accessing raw text and presenting it in a highly usable and intuitive way.

Claims

1. A text processing system, comprising:

a text clustering component configured to cluster sub-document linguistic units of a plurality of relevant documents into clusters;

a sentiment analyzer configured to attribute a sentiment to the linguistic units; and

a report generator configured to identify linguistic units and generate a report based on the clusters and based on the sentiment attributed to the linguistic units.

2. The text processing system of claim 1 and further comprising:

a document identifying engine configured to identify the plurality of relevant documents from among a group of documents larger than the plurality of relevant documents.

3. The text processing system of claim 1 wherein the sentiment analyzer is configured to attribute the sentiment to the linguistic units after the linguistic units are clustered by the text clustering component.

4. The text processing system of claim 3 wherein the sentiment analyzer is configured to calculate a representative sentiment for each cluster.

5. The text processing system of claim 1 wherein the sentiment analyzer is configured to attribute the sentiment to the linguistic units before the linguistic units are clustered.

6. The text processing system of claim 5 wherein the sentiment analyzer is configured to calculate a representative sentiment for each cluster.

7. The text processing system of claim 1 and further comprising:

a display component displaying data indicative of the clusters and the sentiment attributed to the linguistic units.

8. The text processing system of claim 7 and further comprising:

a user feedback component configured to receive a user correction input indicative of a user correction to the data displayed.

9. The text processing system of claim 8 wherein the user feedback component is configured to receive the user correction input as a sentiment correction input indicative of a user correction to sentiment attributed to a linguistic unit.

10. The text processing system of claim 9 wherein the user feedback component is configured to provide the user correction as training data to a machine learning process for training the sentiment analyzer.

11. The text processing system of claim 1 wherein the sentiment analyzer strained using a machine learning process based on features extracted from training data, the features including a set of most salient features.

12. The text processing system of claim 11 wherein the most salient features used in training the sentiment analyzer are removed from use in the clustering engine.

13. The text processing system of claim 11 wherein the sentiment analyzer is trained iteratively on a combination of human-annotated and automatically annotated data.

14. The text processing system of claim 1 wherein the text clustering component comprises a subject matter clustering component configured to cluster the linguistic units based on predefined subject matter.

15. The text processing system of claim 1 wherein the text clustering component is configured to intermittently re-cluster relevant documents including new relevant documents to obtain new clusters.

16. The text processing system of claim 15 wherein the sentiment analyzer is configured to attribute a sentiment to new linguistic units in the new relevant documents.

17. The text processing system of claim 16 wherein the sentiment analyzer is configured to calculate a representative sentiment associated with each new cluster.

18. The text processing system of claim 17 wherein the report generator is configured to report changes in the representative sentiment attributed to the clusters over time.

19. The text processing system of claim 1 and further comprising:

a relevant document data store storing the relevant documents, and wherein the relevant document data store is configured to receive cluster information indicative of the clusters calculated by text clustering component and sentiment information indicative of the sentiment attributed to the linguistic units by the sentiment analyzer.