Voice query refinement to embed context in a voice query

1. A method comprising:

identifying a character in a plurality of frames presented during playback of a video while an ambiguous search query is received;

generating a model of at least one movement of the character that occurred during presentation of at least two frames of the plurality of frames;

comparing the model to at least one movement template in a database;

determining, based on the comparing, the model is comparable to the at least one movement template;

in response to determining the model is comparable to the at least one movement template, extracting a keyword from the at least one movement template; and

retrieving search results for display based on the keyword and the ambiguous search query.

2. The method of claim 1 , wherein extracting the keyword comprises retrieving metadata of the at least one movement template.

3. The method of claim 1 , wherein generating the model comprises:

identifying at least two body parts of the character, wherein each of the at least two body parts are visible in at least two frames of the plurality of frames;

computing at least one angle formed between the at least two body parts; and

storing the at least one angle as part of the model.

4. The method of claim 3 , wherein the at least two body parts are identified as a body part combination.

5. The method of claim 4 , further comprising:

searching the database for the body part combination; and

identifying the at least one movement template based on the body part combination.

6. The method of claim 1 , wherein determining the model is comparable to the at least one movement template comprises determining the model comprises a body part combination angle that matches a corresponding body part combination angle of the at least one movement template.

7. The method of claim 1 , further comprising:

generating context data characterizing objects present in the plurality of frames; and

retrieving the search results based on the context data and the ambiguous search query.

8. The method of claim 1 , further comprising:

receiving a spoken search query;

performing speech to text analysis to derive words that comprise the spoken search query; and

determining at least one word of the spoken search query is ambiguous based on analysis of the derived words.

9. The method of claim 8 , wherein determining the at least one word of the spoken search query is ambiguous comprises determining the at least one word is at least one of a pronoun or an auxiliary verb.

10. The method of claim 1 , wherein the plurality of frames comprises a predetermined number of displayed frames after receiving the ambiguous search query.

11. A system comprising:

input circuitry to receive a query during presentation of a video; and

control circuitry configured to:

identify a character in a plurality of frames presented during playback of a video while an ambiguous search query is received;

generate a model of at least one movement of the character that occurred during presentation of at least two frames of the plurality of frames;

compare the model to at least one movement template in a database;

determine, based on the comparing, the model is comparable to the at least one movement template;

in response to determining the model is comparable to the at least one movement template, extract a keyword from the at least one movement template; and

retrieve search results for display based on the keyword and the ambiguous search query.

12. The system of claim 11 , wherein the control circuitry configured to extract the keyword is further configured to retrieve metadata of the at least one movement template.

13. The system of claim 11 , wherein the control circuitry configured to generate the model is further configured to:

identify at least two body parts of the character, wherein each of the at least two body parts are visible in at least two frames of the plurality of frames;

compute at least one angle formed between the at least two body parts; and

store the at least one angle as part of the model.

14. The system of claim 13 , wherein the at least two body parts are identified as a body part combination.

15. The system of claim 14 , wherein the control circuitry is further configured to:

searching the database for the body part combination; and

identifying the at least one movement template based on the body part combination.

16. The system of claim 11 , wherein the control circuitry configured to determine the model is comparable to the at least one movement template is further configured to determine the model comprises a body part combination angle that matches a corresponding body part combination angle of the at least one movement template.

17. The system of claim 11 , wherein the control circuitry is further configured to:

generate context data characterizing objects present in the plurality of frames; and

retrieve the search results based on the context data and the ambiguous search query.

18. The system of claim 11 , wherein the control circuitry is further configured to:

receive a spoken search query;

perform speech to text analysis to derive words that comprise the spoken search query; and

determine at least one word of the spoken search query is ambiguous based on analysis of the derived words.

19. The system of claim 18 , wherein the control circuitry configured to determine the at least one word of the spoken search query is ambiguous is further configured to determine the at least one word is at least one of a pronoun or an auxiliary verb.

20. The system of claim 11 , wherein the plurality of frames comprises a predetermined number of displayed frames after receiving the ambiguous search query.