Search engine for scientific literature providing interface with automatic image ranking

• US 8,412,703 B2
• Filed: 07/06/2010
• Issued: 04/02/2013
• Est. Priority Date: 07/17/2009
• Status: Expired due to Fees

• Current Assignee: Yu Hong
• Original Assignee: Yu Hong
• Inventors: Yu, Hong
• Primary Examiner(s): Nguyen, Cindy
• Application Number:
  US12/830,839
• Publication Number:
  US 20110016112A1
• Time in Patent Office: 1,001 Days
• Field of Search: 707/723, 707/706
• US Class Current: 707/723
• CPC Class Codes: G06F 16/53 Querying
• CPC Class Codes: G06F 16/583 using metadata automatically derived from the content
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

An information retrieval system for scientific literature provides a summary document for reviewing the results of a query that provides both a text-based summary and an image-based summary, the latter represented by a figure from the document that has a highest ranking indicating its centrality to the concept of the document. A ranking of figures may be done by an analysis of figure-associated text either by supervised machine learning or comparison of text similarities.

Claims

1. An electronic interface for a scientific database query system permitting retrieval from an electronic scientific database of scientific documents having text and associated figures as stored in the electronic scientific database, the figures described at least in part by text, the electronic interface comprising:

an electronic computer executing a stored program to;

(a) receive from the electronic scientific database at least one document selected from the electronic scientific database based on a text query from a human operator; and

(b) review the text of the document and the associated figures to provide a ranking of the figures independent from an ordering of the figures in the document and according to an automated determination of an importance of the figure with respect to a subject matter of the document, wherein the ranking is derived from the text of the document.

2. The electronic interface of claim 1 wherein the electronic computer further executes the stored program to:

(c) create additional searchable data for the document in the electronic scientific database based at least in part on the ranking.

3. The electronic interface of claim 2 wherein the electronic computer further executes the stored program to

(d) produce a display output viewable by the human operator providing a summary document providing a text summary of the text of the document and at least one figure of the document selected according to the ranking of the figures.

4. The electronic interface of claim 3 wherein figures in the summary document are varied in size as a function of the ranking of the figures.

5. The electronic interface of claim 3 wherein the summary document includes multiple figures with lower-ranked figures represented as thumbnail images distinguishable from one another but not of legible sizes.

6. The electronic interface of claim 3 wherein the summary document further includes a text description of at least one figure.

7. The electronic interface of claim 6 wherein the text description is automatically extracted from text of the document other than a caption of the figure.

8. The electronic interface of claim 7 wherein the text description provides at least one sentence selected to address categories of:

introduction, methods, results, and discussion of the document generated from the text of the document.

9. The electronic interface of claim 3 wherein the summary document is attached to the document including a full text and figures of the document, the figures of the summary document duplicated in the full text and figures of the document.

10. The electronic interface of claim 3 wherein the text summary is an abstract portion of the document.

11. The electronic interface of claim 3 further including a feature extractor and a supervised machine learning system, the feature extractor receiving the document and extracting information about the figures from the document and applying the extracted information to the supervised machine learning system, trained using a training set of training documents and ranked figures, to identify a ranking of the figures associated with the document.

12. The electronic interface of claim 11 wherein the supervised machine learning system is selected from the group consisting of:

naï

ve Bayes, decision tree, neural networks, and support vector machines.

13. The electronic interface of claim 11 wherein the supervised machine learning system uses a method selected from the group consisting of logistic regression and conditional random fields.

14. The electronic interface of claim 11 wherein the feature extractor extracts information about the document selected from the group consisting of:

a location of the figure within categories of;

introduction, methods, results, and discussion;

how many references are made to the figure in the text of the document; and

whether the figure is an image-of-thing or a mix image.

15. The electronic interface of claim 3 wherein the ranking of the figure analyzes a similarity of a caption of the figure and other text of the document to identify a ranking of the figures associated with the document.

16. The electronic interface of claim 15 wherein the ranking of the figure analyzes a similarity of a caption of the figure and an abstract of the document.

17. The electronic interface of claim 15 wherein the caption of the figure is augmented with text extracted from the figure by optical character recognition.

18. An method of creating a document summary for a scientific database query system, the query system permitting retrieval from an electronic scientific database of scientific documents having text and associated figures, the figures described at least in part by text, the method implemented using an electronic computer executing a stored program to perform the steps of:

(a) receiving from the electronic scientific database at least one document selected from the electronic scientific database based on a text query from a human operator; and

(b) reviewing the text of the document and the associated figures to provide a ranking of the figures independent from an ordering of the figures in the document and according to an automated determination of an importance of the figure with respect to a subject matter of the document, wherein the ranking is derived from the text of the document.

19. The method of claim 18 further including the step of creating additional searchable data for the document in the electronic scientific database based at least in part on the ranking.

20. The method of claim 18 further including the step of outputting a display viewable by the human operator providing a summary document providing a text summary of the text of the document and at least one figure of the document selected according to the ranking of the figures.