Search engine inference based virtual assistance

• US 9,201,970 B2
• Filed: 06/17/2013
• Issued: 12/01/2015
• Est. Priority Date: 03/16/2010
• Status: Expired due to Fees

• Current Assignee: Streamline Licensing LLC (IP Edge LLC)
• Original Assignee: Empire Technology Development LLC (Allied Inventors Management, LLC)
• Inventors: Kruglick, Ezekiel
• Primary Examiner(s): Boccio, Vincent F
• Application Number:
  US13/919,705
• Publication Number:
  US 20130282692A1
• Time in Patent Office: 897 Days
• Field of Search: 707/706, 707/765, 707/769
• US Class Current: 1/1
• CPC Class Codes: G06F 16/9035 Filtering based on additional data, e.g. user or group profiles
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/957 Browsing optimisation, e.g. caching or content distillation
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Techniques described herein generally relate to real time inference based systems. Example embodiments may set forth devices, methods, and computer programs related to search engine inference based virtual assistance. One example method may include a computing device adapted to receive text as input and a computer processor arranged to determine at least one inference regarding subject matter of the text based on one or more web searches of one or more terms within the text. The inference(s) may then be automatically displayed upon the inference(s) being determined. The text may be automatically received as input from a voice-to-text converter as voice-to-text conversion producing the text is occurring.

Claims

1. A method to display inferences according to evaluated inference strength, comprising:

receiving, by a computing device, streaming audio or video data;

converting, by the computing device, voice within received audio or video data to streaming text;

performing a web search using two or more words within the streaming text as search terms to determine topical relatedness of the two or more words within the streaming text;

when topical relatedness of the two or more words within the streaming text is above a predetermined relatedness threshold, automatically combining, by the computing device, the two or more words within the streaming text with one or more neighboring words within the streaming text to create a larger word combination, and automatically performing a web search using the larger word combination as search terms to determine topical relatedness of the larger word combination; and

when topical relatedness of the two or more words or the larger word combination within the streaming text is below the predetermined relatedness threshold;

automatically performing at least one web search using the two or more words or the larger word combination within the streaming text to identify, by the computing device, at least one inference regarding subject matter of the streaming text;

determining, by the computing device, a strength of the at least one inference; and

displaying the at least one inference by the computing device, along with the streaming text, when the strength of the at least one inference is above a predetermined strength threshold, or displaying the at least one inference by the computing device, along with the streaming text, when the strength of the at least one inference is greater than a strength of one or more other inferences identified by the computing device.

2. The method of claim 1, wherein determining the strength of the at least one inference comprises determining a proportion of web search results, of a web search using the at least one inference as a search term, which web search results include the two or more words or the larger word combination within the streaming text, wherein a higher proportion of web search results including the two or more words or the larger word combination within the streaming text corresponds to a higher inference strength, and wherein a lower proportion of web search results including the two or more words or the larger word combination within the streaming text corresponds to a lower inference strength.

3. The method of claim 1, wherein determining the strength of the at least one inference comprises determining a number of different inferences identified from the at least one web search to identify the at least one inference, wherein a lower number of different inferences corresponds to a higher inference strength, and wherein a higher number of different inferences corresponds to a lower inference strength.

4. The method of claim 1, wherein determining the strength of the at least one inference comprises comparing a proportion of different inferences, identified by the at least one web search to identify the at least one inference, to a number of web search results returned by the at least one web search to identify the at least one inference.

5. The method of claim 1, wherein determining the strength of the at least one inference comprises determining a number of web search results, returned by the at least one web search to identify the at least one inference, the web search results using one or more terms within the streaming text, wherein a lower number of web search results using one or more terms within the streaming text corresponds to a higher inference strength, and wherein a higher number of web search results using one or more terms within the streaming text corresponds to a lower inference strength.

6. The method of claim 1, wherein identifying, by the computing device, the at least one inference comprises selecting one or more rare words within the streaming text, and using the one or more rare words in the at least one web search to identify the at least one inference.

7. The method of claim 6, wherein selecting the one or more rare words within the streaming text comprises comparing a word rarity score to a predetermined threshold value.

8. The method of claim 7, further comprising performing, by the computing device, a word rarity web search and using a number of hits returned from the word rarity web search as a basis for the word rarity score.

9. The method of claim 6, wherein identifying, by the computing device, the at least one inference comprises joining the one or more rare words with neighboring words, and using the one or more rare words and neighboring words in the at least one web search to identify the at least one inference.

10. The method of claim 6, wherein identifying, by the computing device, the at least one inference comprises adding the one or more rare words to an inference term table.

11. The method of claim 1, wherein identifying the at least one inference regarding subject matter of the streaming text further comprises automatically analyzing web search results, returned by the at least one web search to identify the at least one inference, to determine at least one common topic occurring within the web search results, wherein the at least one inference comprises the at least one common topic.

12. A computing device comprising:

a hardware processor;

a memory; and

a search engine inference based virtual assistance component configured to;

receive streaming audio or video data;

convert voice within received audio or video data to streaming text;

perform a web search using two or more words within the streaming text as search terms to determine topical relatedness of the two or more words within the streaming text;

when topical relatedness of the two or more words within the streaming text is above a predetermined relatedness threshold, automatically combine the two or more words within the streaming text with one or more neighboring words within the streaming text to create a larger word combination, and automatically perform a web search using the larger word combination as search terms to determine topical relatedness of the larger word combination; and

when topical relatedness of the two or more words or the larger word combination within the streaming text is below the predetermined relatedness threshold;

automatically perform at least one web search using the two or more words or the larger word combination within the streaming text to identify at least one inference regarding subject matter of the streaming text;

determine a strength of the at least one inference; and

display the at least one inference, along with the streaming text, when the strength of the at least one inference is above a predetermined strength threshold, or display the at least one inference, along with the streaming text, when the strength of the at least one inference is greater than a strength of one or more other inferences identified by the computing device.

13. The computing device of claim 12, wherein the search engine inference based virtual assistance component is configured to determine the strength of the at least one inference by determining a proportion of web search results, of a web search using the at least one inference as a search term, which web search results include the two or more words or the larger word combination within the streaming text, wherein a higher proportion of web search results including the two or more words or the larger word combination within the streaming text corresponds to a higher inference strength, and wherein a lower proportion of web search results including the two or more words or the larger word combination within the streaming text corresponds to a lower inference strength.

14. The computing device of claim 12, wherein the search engine inference based virtual assistance component is configured to determine the strength of the at least one inference by determining a number of different inferences identified from the at least one web search to identify the at least one inference, wherein a lower number of different inferences corresponds to a higher inference strength, and wherein a higher number of different inferences corresponds to a lower inference strength.

15. The computing device of claim 12, wherein the search engine inference based virtual assistance component is configured to determine the strength of the at least one inference by determining a proportion of different inferences, identified by the at least one web search to identify the at least one inference, to a number of web search results returned by the at least one web search to identify the at least one inference.

16. The computing device of claim 12, wherein the search engine inference based virtual assistance component is configured to determine the strength of the at least one inference by determining a number of web search results, returned by the at least one web search to identify the at least one inference, the web search results using one or more terms within the streaming text, wherein a lower number of web search results using one or more terms within the streaming text corresponds to a higher inference strength, and wherein a higher number of web search results using one or more terms within the streaming text corresponds to a lower inference strength.

17. The computing device of claim 12, wherein the search engine inference based virtual assistance component is configured to identify the at least one inference by selecting one or more rare words within the streaming text, and using the one or more rare words in the at least one web search to identify the at least one inference.

18. The computing device of claim 12, wherein identifying the at least one inference regarding subject matter of the streaming text further comprises automatically analyzing web search results, returned by the at least one web search to identify the at least one inference, to determine at least one common topic occurring within the web search results, wherein the at least one inference comprises the at least one common topic.

19. A non-transitory computer readable storage medium having computer executable instructions executable by a processor, the instructions that, when executed by the processor, cause the processor to:

receive streaming audio or video data;

convert voice within received audio or video data to streaming text;

perform a web search using two or more words within the streaming text as search terms to determine topical relatedness of the two or more words within the streaming text;

when topical relatedness of the two or more words within the streaming text is above a predetermined relatedness threshold, automatically combine the two or more words within the streaming text with one or more neighboring words within the streaming text to create a larger word combination, and automatically perform a web search using the larger word combination as search terms to determine topical relatedness of the larger word combination; and

when topical relatedness of the two or more words or the larger word combination within the streaming text is below the predetermined relatedness threshold;

automatically perform at least one web search using the two or more words or the larger word combination within the streaming text to identify at least one inference regarding subject matter of the streaming text;

determine a strength of the at least one inference; and

display the at least one inference, along with the streaming text, when the strength of the at least one inference is above a predetermined strength threshold, or display the at least one inference, along with the streaming text, when the strength of the at least one inference is greater than a strength of one or more other inferences identified by the computing device.