Abstract: User intent may be inferred from mouse movements made within a user interface. Client-side instrumentation may be provided that collects mouse movement data that is provided to a classification engine. The classification engine receives the mouse movement data and creates a mouse trajectory. The mouse trajectory may be split into segments, and features associated with each segment may be determined. Features representing the context of the search, that is, content of the search result page, previous queries submitted, and interaction features such as scrolling, may be included. By examining the features associated with the mouse trajectories within the context of a search session, the user intent may be classified into categories using machine learning classification techniques. By inferring user intent, Web search engines may be able to predict whether a user's intent is commercial and tailor advertising accordingly.

Abstract: The present invention is directed towards systems and methods for identifying high quality content in a social media environment. The method according to one embodiment of the present invention comprises retrieving a content item and retrieving a plurality of quality features associated with said content item wherein said quality features comprise intrinsic, usage and relationship features. The method then performs an analysis of said content item against said quality features and generates a quality score based on said analysis.

Abstract: An system for segmenting strings into component parts for use with a database management system. A reference table of string records are segmented into multiple substrings corresponding to database attributes. The substrings within an attribute are analyzed to provide a state model that assumes a beginning, a middle and an ending token topology for that attribute. A null token takes into account an empty attribute component and copying of states allows for erroneous token insertions and misordering. Once the model is created from the clean data, the process breaks or parses an input record into a sequence of tokens. The process then determines a most probable segmentation of the input record by comparing the tokens of the input record with a state models derived for attributes from the reference table.

Abstract: The invention is a method for retrieving answers to questions from an information retrieval system. The method involves automatically learning phrase features for classifying questions into different types, automatically generating candidate query transformations from a training set of question/answer pairs, and automatically evaluating the candidate transforms on information retrieval systems. At run time, questions are transformed into a set of queries, and re-ranking is performed on the documents retrieved.

Abstract: Systems and methods that estimate user preference, via automatic interpretation of user behavior. A user behavior component associated with a search engine can automatically interpret collective behavior of users (e.g., web search users). Such feedback component can include user behavior features and predictive models (e.g., from a user behavior component) that are robust to noise, which can be present in observed user interactions with the search results (e.g., malicious and/or irrational user activity.

Abstract: Structured content and associated metadata from the Web are leveraged to provide specific answer string responses to user questions. The structured content can also be indexed at crawl-time to facilitate searching of the content at search-time. Ranking techniques can also be employed to facilitate in providing an optimum answer string and/or a top K list of answer strings for a query. Ranking can be based on trainable algorithms that utilize feature vectors for candidate answer strings. In one instance, at crawl-time, structured content is indexed and automatically associated with metadata relating to the structured content and the source web page. At search-time, candidate indexed structured content is then utilized to extract an appropriate answer string in response to a user query.

Abstract: The subject disclosure pertains to systems and methods for training machine learning systems. Many cost functions are not smooth or differentiable and cannot easily be used during training of a machine learning system. The machine learning system can include a set of estimated gradients based at least in part upon the ranked or sorted results generated by the learning system. The estimated gradients can be selected to reflect the requirements of a cost function and utilized instead of the cost function to determine or modify the parameters of the learning system during training of the learning system.

Abstract: An system for segmenting strings into component parts for use with a database management system. A reference table of string records are segmented into multiple substrings corresponding to database attributes. The substrings within an attribute are analyzed to provide a state model that assumes a beginning, a middle and an ending token topology for that attribute. A null token takes into account an empty attribute component and copying of states allows for erroneous token insertions and misordering. Once the model is created from the clean data, the process breaks or parses an input record into a sequence of tokens. The process then determines a most probable segmentation of the input record by comparing the tokens of the input record with a state models derived for attributes from the reference table.

Abstract: The invention is a method for retrieving answers to questions from an information retrieval system. The method involves automatically learning phrase features for classifying questions into different types, automatically generating candidate query transformations from a training set of question/answer pairs, and automatically evaluating the candidate transforms on information retrieval systems. At run time, questions are transformed into a set of queries, and re-ranking is performed on the documents retrieved.