Media content search engine incorporating text content and user log mining

A method comprising:

• identifying a media content source;
• collecting one or more pieces of media content and associated text from the media content source;
• extracting, for a piece of media content, one or more text features from the associated text;
• and
• making the one or more text features available for searching.

A method comprising:

• receiving search criteria;
• generating a query vector based on the search criteria;
• comparing the query vector to a feature vector corresponding to a piece of media content and having been generated based on text associated with the piece of media content;
• and
• determining, based at least in part on a result of the comparing, whether to render the piece of media content to a user.

A system comprising:

• a crawler module coupled to access a media content source and collect a plurality of media content pieces and associated text from the media content source;
• a feature extraction module coupled to extract one or more text features from one of the media content pieces;
• and
• a media content indexing module coupled to generate a text feature vector, based on the extracted one or more text features, corresponding to the one media content piece.

A system comprising:

• a query generator to generate a query vector based on received search criteria;
• and
• a matching module coupled to,
• receive the query vector and compare the query vector to a plurality of feature vectors corresponding to a plurality of pieces of media content, wherein each of the plurality of feature vectors has been generated based on text associated with one of the plurality of pieces of media content, and
• identify one or more of the plurality of pieces of media content to return for rendering to a user based on the comparison of the query vector to the plurality of feature vectors.

A method comprising:

• receiving search criteria;
• identifying, based at least in part on the search criteria, a piece of media content to be rendered;
• receiving user feedback regarding the relevancy of the rendered piece of media content;
• weighting for another piece of media content, based on the user feedback, both a result of comparing the high-level query vector to a high-level feature vector of the other piece of media content and a result of comparing the low-level query vector to a low-level feature vector of the other piece of media content;
• and
• combining the weighted result to determine whether to identify the other piece of media content for rendering.

A method comprising:

• receiving user feedback regarding the relevancy of one or more pieces of rendered media content;
• and
• for each of the one or more pieces of media content,
• modifying, based on the user feedback, a feature vector corresponding to the piece of media content, and
• making the modified feature vector available for subsequent searching of the one or more pieces of media content.

One or more computer-readable media having stored thereon a plurality of instructions that, when executed by one or more processors of a computer, causes the one or more processors to perform acts including:

• identifying a piece of media content to render to a user based at least in part on comparing a query vector corresponding to search criteria of the user and a feature vector corresponding to the piece of media content;
• receiving user feedback regarding the relevancy of the piece of media content;
• modifying the query vector based on the received user feedback;
• and
• modifying the feature vector based on the received user feedback in an off-line log mining process.