Search engine for textual content and non-textual content

• US 10,445,367 B2
• Filed: 05/14/2013
• Issued: 10/15/2019
• Est. Priority Date: 05/14/2013
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Wallenberg, Mattias, Arngren, Tommy
• Primary Examiner(s): Hu, Jensen
• Application Number:
  US14/891,259
• Publication Number:
  US 20160085860A1
• Time in Patent Office: 2,345 Days
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/73 Querying
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9538 Presentation of query results

Abstract

A search engine system that can match a search request to not only a specific content item (e.g., video file), but also to a single component of a content item. For instance, using a video content item as an example, the search engine system can match a specific search request to not only a specific video within a collection of videos, but also to a single moment within a video, a video segment, and a group of videos.

Claims

1. A method performed by a search engine system (SES), the method comprising:

receiving, at the SES, a search request transmitted by a client device, wherein said search request includes one or more query terms;

determining, by the SES, a query vector based on said one or more query terms;

determining, by the SES, a first set of tag vectors for a first set of tags associated with a first segment of a first non-textual content item;

determining, by the SES, a second set of tag vectors for a second set of tags associated with a second segment of said first non-textual content item;

determining, by the SES, a first segment vector for said first segment by summing said first set of tag vectors;

determining, by the SES, a second segment vector for said second segment by summing said second set of tag vectors;

calculating, by the SES, a first segment search score based on a result of a comparison of said first segment vector to said query vector;

calculating, by the SES, a second segment search score based on a result of a comparison of said second segment vector to said query vector; and

comparing said first segment search score and said second segment search score,

wherein

one or more vectors of the first and second sets of tag vectors is a weighted tag vector,

the weighted tag vector is obtained by multiplying an initial tag vector with a feature score,

the feature score is determined based on a feature type of a tag, and

the feature type is one of image, audio, video, and text.

2. The method of claim 1, further comprising:

determining, by the SES, a first item vector for said first non-textual content item;

determining, by the SES, a second item vector for a second non-textual content item;

determining, by the SES, a first item search score, wherein said first item search score is based on a comparison of said first item vector to said query vector;

determining, by the SES, a second item search score, wherein said second item search score is based on a comparison of said second item vector to said query vector; and

selecting one or more of;

said first segment, said second segment, said first non-textual content item, and said second non-textual content item based on said first segment search score, second segment search score, said first item search score, and said second item search score.

3. The method of claim 1, further comprising:

transmitting an ordered set of two or more search results based on the search request, wherein said ordered set of search results includes a first search result that comprises information identifying said first segment, wherein the position of said first search result within said ordered set of search results is determined based on said first segment search score and a search score associated with each search result included in said ordered set of search results.

4. A method performed by a search engine system (SES), the method comprising:

receiving, at the SES, a search request transmitted by a client device, wherein said search request includes one or more query terms;

determining, by the SES, a query vector based on said one or more query terms;

determining, by the SES, a first weighted tag vector based on said one or more query terms and a first tag, wherein said first tag is linked with a first feature located in a first segment of a non-textual content item;

determining, by the SES, a second weighted tag vector based on said one or more query terms and a second tag, wherein said second tag is linked with a second feature located in a second segment of the non-textual content item;

calculating, by the SES, a first tag search score based on a result of a comparison of said first weighted tag vector to said query vector; and

calculating, by the SES, a second tag search score based on a result of a comparison of said second weighted tag vector to said query vector, wherein

said first weighted tag vector is obtained by multiplying a first initial tag vector with a feature score;

the feature score is determined based on a feature type of said first tag, and

the feature type is one of image, audio, video, and text.

5. The method of claim 4, further comprising:

determining, by the SES, an item collection vector for a group of non-textual content items including said non-textual content item.

6. The method of claim 4, wherein said search request includes a search type indicator that indicates a user is requesting, at the least, a tag search.

7. A search engine system (SES) comprising:

a data storage system and a data processing system, said data storage system comprising instructions executable by the data processing system whereby the SES is operative to;

determine a query vector based on query terms included in a search request;

determine a first set of tag vectors for a first set of tags associated with a first segment of a first non-textual content item;

determine a second set of tag vectors for a second set of tags associated with a second segment of said first non-textual content item;

determine a first segment vector for said first segment by summing said first set of tag vectors;

determine a second segment vector for said second segment by summing said second set of tag vectors;

calculate a first segment search score based on a result of a comparison of said first segment vector to said query vector;

calculate a second segment search score based on a result of a comparison of said second segment vector to said query vector; and

compare said first segment search score and said second segment search score, wherein the SES is operative to;

calculate said first segment search score by, at least, calculating;

(VQ·

VS1)/(∥

VQ∥

∥

VS1∥

), where VQ is said query vector, and VS1 is said first segment vector, and

calculate said second segment search score by, at least, calculating;

(VQ·

VS2)/(∥

VQ∥

∥

VS2∥

), where VS2 is said second segment vector.

8. The SES of claim 7, wherein

one or more vectors of the first and second sets of tag vectors is a weighted tag vector,

the weighted tag vector is obtained by multiplying an initial tag vector with a feature score,

the feature score is determined based on a feature type of a tag, and

the feature type is one of image, audio, video, and text.

9. The SES of claim 7, wherein the SES is operative to:

determine a first item vector for said first non-textual content item;

determine a second item vector for a second non-textual content item;

determine a first item search score, wherein said first item search score is based on a comparison of said first item vector to said query vector; and

determine a second item search score, wherein said second item search score is based on a comparison of said second item vector to said query vector.

10. The SES of claim 9, wherein the SES is operative to:

select one or more of;

said first segment, said second segment, said first non-textual content item, and said second non-textual content item based on said first segment search score, said second segment search score, said first item search score, and said second item search score.

11. The SES of claim 7, wherein the SES is operative to:

determine an item collection vector for a group of non-textual content items including said first non-textual content item.

12. The SES of claim 7, wherein the SES is operative to:

transmit an ordered set of two or more search results based on the search request, wherein said ordered set of search results includes a first search result that comprises information identifying said first segment, wherein a position of said first search result within said ordered set of search results is determined based on said first segment search score and a search score associated with each search result included in said ordered set of search results.

13. The SES of claim 7, wherein said search request includes a search type indicator that indicates a user is requesting a segment search.

14. A search engine system (SES) comprising:

a data storage system and a data processing system, said data storage system comprising instructions executable by the data processing system whereby the SES is operative to;

determine a query vector based on one or more query terms included in a search request;

determine a first weighted tag vector based on said one or more query terms and a first tag, wherein said first tag is linked with a first feature located in a first segment of a first non-textual content item;

determine a second weighted tag vector based on said one or more query terms and a second tag, wherein said second tag is linked with a second feature located in a second segment of said first non-textual content item;

calculate a first tag search score based on a result of a comparison of said first weighted tag vector to said query vector; and

calculate a second tag search score based on a result of a comparison of said second weighted tag vector to said query vector, wherein

said first weighted tag vector is obtained by multiplying a first initial tag vector with a feature score,

the feature score is determined based on a feature type of said first tag, and

the feature type is one of image, audio, video, and text, wherein

the SES is operative to;

calculate said first tag search score by, at least, calculating;

(VQ·

VT1)/(∥

VQ∥

∥

VT1∥

), where VQ is said query vector, and VT1 is said first weighted tag vector, and

determine said second tag search score by, at least, calculating;

(VQ·

VT2)/(∥

VQ∥

∥

VT2∥

), where VT2 is said second weighted tag vector.

15. The SES of claim 14, wherein the SES is operative to:

determine a first item vector for said first non-textual content item;

determine a second item vector for a second non-textual content item;

determine a first item search score, wherein said first item search score is based on a comparison of said first item vector to said query vector; and

determine a second item search score, wherein said second item search score is based on a comparison of said second item vector to said query vector.

16. The SES of claim 15, wherein the SES is operative to:

select one or more of;

said first tag, said second tag, said first non-textual content item, and said second non-textual content item based on said first tag search score, said second tag search score, said first item search score, and said second item search score.

17. The SES of claim 14, wherein the SES is operative to:

determine an item collection vector for a group of non-textual content items including said first non-textual content item.

18. The SES of claim 14, wherein the SES is operative to:

transmit an ordered set of two or more search results based on the search request, wherein said ordered set of search results includes a first search result that comprises information identifying said first tag, wherein a position of said first search result within said ordered set of search results is determined based on said first tag search score and a search score associated with each search result included in said ordered set of search results.

19. The SES of claim 14, wherein said search request includes a search type indicator that indicates a user is requesting a tag search.

20. A computer program product comprising a non-transitory computer readable medium storing computer instructions for searching content, the computer instructions comprising:

instructions for determining a query vector based on query terms included in a search request;

instructions for determining a first set of tag vectors for a first set of tags associated with a first segment of a non-textual content item;

instructions for determining a second set of tag vectors for a second set of tags associated with a second segment of said non-textual content item;

instructions for determining a first segment vector for said first segment by summing said first set of tag vectors;

instructions for determining a second segment vector for said second segment by summing said second set of tag vectors;

instructions for calculating a first segment search score based on a result of a comparison of said first segment vector to said query vector;

instructions for calculating a second segment search score based on a result of a comparison of said second segment vector to said query vector; and

instructions for comparing said first segment search score and said second segment search score, wherein

one or more vectors of the first and second sets of tag vectors is a weighted tag vector,

the weighted tag vector is obtained by multiplying an initial tag vector with a feature score,

the feature score is determined based on a feature type of a tag, and

the feature type is one of image, audio, video, and text.

21. A computer program product comprising a non-transitory computer readable medium storing computer instructions for searching content, the computer instructions comprising:

instructions for determining a query vector based on one or more query terms included in a search request;

instructions for determining a first weighted tag vector based on said one or more query terms and a first tag, wherein said first tag is linked with a first feature located in a first segment of a non-textual content item;

instructions for determining a second weighted tag vector based on said one or more query terms and a second tag, wherein said second tag is linked with a second feature located in a second segment of the non-textual content item;

instructions for calculating a first tag search score based on a result of a comparison of said first weighted tag vector to said query vector; and

instructions for calculating a second tag search score based on a result of a comparison of said second weighted tag vector to said query vector, wherein

said first weighted tag vector is obtained by multiplying a first initial tag vector with a feature score,

the feature score is determined based on a feature type of said first tag, and

the feature type is one of image, audio, video, and text.