Microsoft Technology Licensing, LLC

Nicholas Eric Craswell

Hugh Evan Williams

Ariel J Lazier

David Anthony Hawking

Paul Nathan Bennett

Nitin Agrawal

Tabreez Govani

Michael Maxwell Cameron

Chenyan Xiong

Saurabh Kumar Tiwary

Xia Song

Martin Szummer

Peter Richard Bailey

Corbin Louis Rosset

Ryen William White

William Ramsey

Benoit F Dumoulin

Michael James Taylor

John Guiver

Nikhil Dandekar

Girish Kumar

Fernando David Diaz

Onno R Zoeter

Jamie Paul Buckley

Mikhail Bilenko

Mark Blelock Atherton

Amir Jonatan Padovitz

Srinath Reddy Aaleti

Krishna Chaitanya Gade

Amit Prakash

Daniel Fernando Campos

Bodo Von Billerbeck

Andy Lam

Amit Aggarwal

Edward Lloyd Snelson

Bhaskar Mitra

Corby Louis Rosset

Song Zhou

Emine Yilmaz

Hongfei Zhang

Alfian Tan

A computer-implemented technique is described herein for assisting a user in advancing a task objective. The technique uses a suggestion-generating system (SGS) to provide one or more suggestions to a user in response to at least a last-submitted query provided by the user. The SGS may correspond to a classification-type or generative-type neural network. The SGS uses a machine-trained model that is trained using a multi-task training framework based on plural groups of training examples, which, in turn, are produced using different respective example-generating methods. One such example-generating method constructs a training example from queries in a search session. It operates by identifying the task-related intent the queries, and then identifying at least one sequence of queries in the search session that exhibits a coherent task-related intent. A training example is constructed based on queries in such a sequence.

Using a multi-task-trained neural network to guide interaction with a query-processing system via useful suggestions

Described herein is a mechanism for utilizing a neural network to identify and rank search results. A machine learning model is trained by converting training data comprising query-document entries into query term-document entries. The query term-document entries are utilized to train the machine learning model. A set of query terms are identified. The query terms can be derived from a query history. The trained machine learning model is used to calculate document ranking scores for the query terms and the resultant scores are stored in a pre-calculated term-document index. A query to search the document index is broken down into its constituent terms and an aggregate document ranking score is calculated from a weighted sum of the document ranking scores corresponding to the individual query terms. Because the term-document index can be pre-calculated, it can be downloaded to provide deep learning search capabilities in a computationally limited environment.

Neural network for search retrieval and ranking

Representative embodiments disclose mechanisms to complete partial queries entered by a user. Users enter a partial query. The partial query is used to search a short text index comprising the titles of documents. The search yields a list results. The top k entries of the list are selected and a language model is created from the top k entries. The language model comprises n-grams from the top k entries and an associated probability for each n-gram. A query completion generator creates query completion suggestions by matching n-grams with the partial query, removing candidate suggestions that to not comply with suggestion rules, and filtering the remaining suggestions according to a filtering criteria. The top N results are returned as suggestions to complete the query.

Corpus specific generative query completion assistant

A computer-implemented technique receives an input expression that a user submits with an intent to accomplish some objective. The technique then uses a machine-trained intent encoder component to map the input expression into an input expression intent vector (IEIV). The IEIV corresponds to a distributed representation of the intent associated with the input expression, within a vector intent vector space. The technique then leverages the intent vector to facilitate some downstream application task, such as the retrieval of information. Some application tasks also use a neighbor search component to find expressions that express an intent similar to that of the input expression. A training system trains the intent encoder component based on the nexus between queries and user clicks, as recorded in a search engine's search log.

Intent encoder trained using search logs

Non-limiting examples of the present disclosure describe low latency pre-web classification of query data. In examples, processing is performed where query data may be analyzed in a low latency manner that includes providing a vertical intent classification and entity identification for query data before a web ranking service processes the query data. Query data may be received. A vertical intent classification index may be searched using the query data. In examples, the vertical intent classification index may comprise a set of files that can be used to determine one or more candidate entity identifiers for the query data. The one or more entity identifiers may be ranked. The query data, a vertical intent classification for the vertical intent classification index and the one or more ranked candidate entity identifiers may be transmitted for processing associated with a web ranking service. Other examples are also described.

Low latency pre-web classification

Embodiment described herein are directed to an enhanced search engine with multiple feedback loops for providing optimal search results that are responsive a user's search query. The user's search query is parsed, and based on the underlying terms, different linguistic models and refinement techniques generate alternative candidate search queries that may yield better results. Searches are performed for the original search query and the candidate search queries, and different scores are used to select the best search results to present to the user. Results making it onto the list, as well as the underlying candidate search query, linguistic model, or refinement technique for generating that search query, will have their corresponding scores updated to reflect their success of generating a search result. Scores are stored and used by future searches to come up with better results.

Self-tuning alterations framework

Subject matter described herein is related to determining a document score, which suggests a relevance of a document (e.g., webpage) to a search query. For example, a search query is received that is comprised of one or more terms, which represent a subject. An equivalent subject is identified that is semantically similar to the subject. The document score is determined by accounting for both a subject frequency and an equivalent-subject frequency.

Information retrieval using subject-aware document ranker

Computer-readable media and computerized methods for identifying candidate points on a graphical depiction of relative popularity of an entity (e.g., entertainer, sports team, and the like) are provided. Points on the graphical depiction are ranked based on a number of user-submitted web queries that reference the entity that are received during a particular time frame. Peak points and slope values (i.e., derived from an angle of inclination of inclines on the graphical depiction) may be captured by analyzing movements in the rank of an entity over time. An algorithmic process may then be applied to the peak points and slope values to determine points of interest of the entity's popularity, such as the highest-ranked periods and/or dramatic positive movements in rank of the entity. These points of interest are selected as candidate points and are surfaced as icons on a visual representation of the graphical depiction.

Identifying on a graphical depiction candidate points and top-moving queries

A computer system, method, and media for associating locations with ranked websites are provided. The computer system includes a search engine, a log database, and a location database that are employed to respond to search requests from users by returning appropriately ranked websites to the user. The websites are ranked using the location of the website and the location of the user to select websites to receive high ranks. Additionally, the search engine includes a correction feature that reevaluates locations for a website or user when a large number of obtained locations suggest a different location than a currently associated location for the website or the user.

Promoting websites based on location

An approach is provided for determining related queries for a given search query based on the linking structure of electronic documents within a document set. Document titles are used to represent potential search queries and links between the electronic documents are used to determine relationships between the potential search queries. As such, the document set may be represented as a directed graph in which document titles (which represent potential search queries) are nodes and links are edges between the nodes. When a particular search query is received, a corresponding node is identified and related queries are determined by identifying other nodes having connections with that node.

Using link structure for suggesting related queries

A method, system, and medium are provided for enhancing a search result snippet that is associated with a URL. Potentially descriptive terms of the URL are identified by reconciling multiple sources of terms related to the URL. One aspect includes comparing search terms of historically relevant queries to anchor text for the URL. The search result snippet might be intelligently enhanced by causing common terms to be visibly distinguished when the snippet is displayed.

Intelligent enhancement of a search result snippet

An inversion of the basic format of searching is provided herein. Instead of receiving a search query and providing web page results, a search engine receives a web page identifier as search input from an end user, determines related search queries for the associated web page, and provides the related search queries to the end user issuing the search. Related search queries for web pages may also be used to refine search engines performing the basic form of searching by facilitating the determination of web pages to index and the ranking of web pages as search results to user queries. Additionally, related search queries may be used in advertising revenue generation and sharing.

Related search queries for a webpage and their applications

An information retrieval system is described for retrieving a list of documents such as web pages or other items from a document index in response to a user query. In an embodiment a prediction engine is used to predict both explicit relevance information such as judgment labels and implicit relevance information such as click data. In an embodiment the predicted relevance information is applied to a stored utility function that describes user satisfaction with a search session. This produces utility scores for proposed lists of documents. Using the utility scores one of the lists of documents is selected. In this way different sources of relevance information are combined into a single information retrieval system in a principled and effective manner which gives improved performance.

Information retrieval system

Information retrieval using query-document pair information is described. In an embodiment, a click record is accessed having information about queries and documents where user clicks have been observed for query-document pairs. A click graph is either formed or accessed. This has nodes connected by edges, each node representing any of a document and a query and each edge representing at least one observed click. Given at least one first node in the click graph, a similarity measure is determined between that first node and each of one or more second nodes. The second nodes are then ranked on the basis of the similarity measure results and the ranking is used to retrieve information from the click record.

Information retrieval using query-document pair information

Growing community of inventors

Cambridge, United Kingdom

Nicholas Eric Craswell