Oracle International Corporation

Fortytwo, Inc.

Linkedin Corporation

Confluent, Inc.

Yasuhiro Matsuda

Eishay Smith

Danny Blumenfeld

John JianShi Wang

Anna Povzner

Wesley Lin

Mohammad Faisal

Chanh Nguyen

Esteban Kozak

Alejandro Crosa

Dan Norwood

Xiaoyang Gu

Neha Narkhede

Hao Yan

Joseph Adler

Jay Kreps

Geeta Arora

Sundeep Abraham

Garrett Kaminaga

Rikin Gandhi

A stream verification system for a distributed message queue system with metric collectors on each producer and consumer. A producer time stamp allows correlation of sent and received messages. Verification reports are organized by a message topic. A cumulative checksum allows detection of missing or corrupted messages. Verification messages are used to determine if a zero message report means no messages were sent, or rather that the messages weren't received.

Distributed message queue stream verification

A network update interface is presented to a user on a network to display network updates from other users of a mutual social-networking site. The network updates shared by the other users are gathered in a stream and supplied to a facet-filtering system including a network update interface. The user controls the display of certain network update items according to facet-filter characteristics enabled in facet-filter selection panels in the network update interface. The facet-filter characteristics are used by a facet filter to select certain network updates for display to the user in the network update interface. Trending links to further articles with content corresponding to the facet-filter characteristics are displayed to the user according to greatest popularity among the other users. Links to the profiles of the users sharing the articles are also provided in the network update interface.

Facet-based filtering of social network update data

A semantic vector is generated for a search term based upon a global frequency of other, closely related terms within a corpus that is used to compute the semantic vector relative to the search term. The semantic vector is used in connection with a textual search engine, responsive to a user query comprising a search term, to promote any of documents and sites within results returned to the query by the search engine that contain other, closely related terms that strongly correlate with the search term.

Semantic vector in a method and apparatus for keeping and finding information

A find engine provides a user with access to information via a find interface and a keep interface. The find interface accesses both a content search index and a personal search index; the keep interface accesses personal search index. The content search index lists document content, user generated content, and semantic vectors; the personal search index lists user generated content, user explicit gestures, and user implicit gestures. The user explicit gestures comprise keeps that are captured via the keep interface and that consist of any of one or more documents and sites from among the search results to which the user ascribes significance by affirmatively keeping the document or site. The keeps are preserved as one or more persistent search objects and the search results returned to the user are ranked and/or returned in response to the query at least in part in accordance with the keeps.

Persistent search object in a method and apparatus for keeping and finding information

An expert network in connection with keeping and finding information provides a textual search engine which, in response to a user query comprising a search term, uses a semantic vector to promote documents that contain other, closely related terms that strongly correlate with the search term. A social graph is generated for the user in which connections for the user comprise hyper-dimensional relationships based on semantic vectors that link the search term with a collection of the other, closely related terms. A personalized semantic vector for the user is applied to the social graph. The expert network applies the user's personalized semantic vector to locate experts in the user's social graph based on a user query topic. When a connection is considered an expert for the user, content that the connection kept is recommended to the user.

Expert network in a method and apparatus for keeping and finding information

Systems, methods, and other embodiments associated with selectively acquiring and utilizing query statistics are described. One example method includes selectively acquiring statistics associated with search queries submitted to a search process during a sample period. The statistics may be associated with an index and tokens associated with the index and search queries. The method may also include selecting tokens and/or queries to optimize based, at least in part, on the statistics. In one example, whether and/or how the tokens and/or queries are optimized may depend on estimating index fragmentation based, at least in part, on the statistics.

Query statistics

Systems, methodologies, media, and other embodiments associated with efficiently computing document similarity are described. One exemplary system embodiment includes logic to produce a gram from a string and logic to identify candidate documents based on identifying matches between query grams and document grams stored in an inverted index that relates grams to documents. The example system may also include logic to selectively partially reconstruct a candidate document from entries in the inverted index and logic to compute an edit distance between a string associated with a query and a string associated with the partially reconstructed candidate document. The example system may also include a signal logic configured to provide a signal corresponding to the edit distance.

Efficient computation of document similarity

A method and apparatus for processing a query is provided. A query for a set of Matching XML documents is received. A client may transmit the query for a set of XML documents to the database server. The database server transmits a request for the set of XML documents to a persistent storage. The persistent storage uses an index, at least in part, to obtain the set of XML documents sought by the query. The index stores a first set of entries comprising information about the location of elements, a second set of entries comprising information about element attributes, and a third set of entries comprising information about element values. In responding to a particular query, additional XML documents that are not indexed by the index may also be checked to determine if they correspond to the set of search criteria.

Method and mechanism for processing queries for XML documents using an index

A system and method is provided for generating an inverted index and processing search queries using the inverted index. To increase efficiency for queries having multiple numeric range conditions, numeric attributes are tokenized into a plurality of tokens based on their binary value. The tokens become keys in the inverted index. A numeric range query is translated into a query on multiple tokens and combining two or more range queries on different attributes becomes a simple merge of document identification lists.

Inverted index system and method for numeric attributes

A compressor receives a plurality of characters in a character string, and encodes the input character string to generate an encoded string. The encoding preserves the original binary order of the input character strings in the encoded strings. A predicted character is generated for each character based on prior character sequences in the character string. In one embodiment, a hash based predictive technique is used to generate the predicted characters. A correct order indicating token is generated for an input character if a predicted character, corresponding to the input character, is the input character. The compressor stores the order indicating token in the encoded string to represent the character. However, if an input character has a binary representation greater than a binary representation of a corresponding predicted character, then a failure greater than token and the input character are stored in the encoded string to represent the input character. A failure less than token and an input character are stored for an input character in the encoded string if the input character has a binary representation less than a binary representation of the predicted character. Thus, binary order is preserved in the encoded string. With the binary order preserving compression technique, data may be sorted while the data are still compressed. Prefix bits compression and dynamic token generation are also disclosed.

Growing community of inventors

Palo Alto, CA, United States of America

Yasuhiro Matsuda