Yahoo! Inc.

Excalibur Ip, LLC

Aristides Gionis

Francesco Bonchi

Michele Berlingerio

Flavio Junqueira

Ricardo Alberto Baeza-Yates

Vassilis Plachouras

Debora Donato

Björn Bringmann

Paolo Ferragina

Barbara Poblete

Ingmar Weber

Marco Serafini

Claudio Corsi

Mauro Sozio

Luca Telloli

Flavio Junqueria

Gianmarco De Francisci Morales

A network's evolution is characterized by graph evolution rules. A graph, formed by merging multiple graphs representing the multiple snapshots of the network, that represents an evolutionary network is mined to identify evolutional patterns of the network. A pattern is selected from the identified patterns. Graph evolution rules are generated using identified evolutional patterns. The generated graph evolution rules represent the evolutional patterns of the network, the rules indicating that any occurrence of a child pattern of the selected pattern implies a corresponding occurrence of the selected pattern.

Network graph evolution rule generation

Disclosed are methods and apparatus for detecting spam hosts. In one embodiment, one or more graphs are generated using data obtained from a query log, where the one or more graphs include at least one of an anticlick graph or a view graph. Values of one or more syntactic features of the one or more graphs are ascertained. Values of one or more semantic features of the one or more graphs are determined by propagating categories from a web directory among nodes in each of the one or more graphs. Spam hosts are then detected based upon the values of the syntactic features and the semantic features.

Query log mining for detecting spam hosts

Disclosed herein are a system, method and architecture for optimizing a manner in which queries and updates are issued to data stores servicing an application such as a social networking application. A social graph modeling a social network may be used to generate a data store request schedule that optimizes, e.g., minimizes, data store accesses, such as without limitation queries and updates to the data store.

Social networking feed delivery system and method

A matching between content items and consumers is discloses. More particularly, items and consumers are matched using a matching approach that uses capacity constraints associated with each consumer, capacity constraints associated with each item, and relationship weights, each relationship weight representing a similarity between a consumer and an item. Edges representing the relationships between consumers and items can be selected using an iterative selection that includes a matching approach that permits capacity constraints. Alternatively, edges can be selected using an iterative approach that allows a solution to be identified prior to completion of the selection processing.

System and method of matching content items and consumers

Methods and apparatus are described for modeling sequences of events with Markov models whose states correspond to nodes in a provided taxonomy. Each state represents the events in the subtree under the corresponding node. By lumping observed events into states that correspond to internal nodes in the taxonomy, more compact models are achieved that are easier to understand and visualize, at the expense of a decrease in the data likelihood. The decision for selecting the best model is taken on the basis of two competing goals: maximizing the data likelihood, while minimizing the model complexity (i.e., the number of states).

Taxonomy-driven lumping for sequence mining

Techniques for query processing in a multi-site search engine are described. During an indexing phase, each site of a multi-site search engine indexes a set of assigned web resources and each site calculates, for each term in the set of assigned web resources, a site-specific upper bound ranking score on the contribution of the term to the search engine ranking function for a query containing the term. During a propagation phase, all sites exchange their site-specific upper bound ranking scores with each other. In response to a site receiving a query, the site determines the set of locally matching resources and compares the ranking score of a locally matching resource with the site-specific upper bound ranking scores for the terms of the query that were received during the propagation phase and determines whether to communicate the query to other sites. By exchanging appropriately defined site-specific upper bound ranking scores, the site initially receiving the query can determine whether the locally matching resources would be identical to the resources obtained from a single-site search system without having to communicate the query to each of the other sites.

System and methodology for a multi-site search engine

A method of ascribing scores to web documents and search queries generates a hyperlink-click graph by taking the union of the hyperlink and click graphs, takes a random walk on the hyperlink-click graph, and associates the transition probabilities resulting from the random walk with scores for each of the documents and search queries.

Method and system for generating a hyperlink-click graph

A method of caching posting lists to a search engine cache calculates the ratios between the frequencies of the query terms in a past query log and the sizes of the posting lists for each term, and uses these ratios to determine which posting lists should be cached by sorting the ratios in decreasing order and storing to the cache those posting lists corresponding to the highest ratio values. Further, a method of finding an optimal allocation between two parts of a search engine cache evaluates a past query stream based on a relationship between various properties of the stream and the total size of the cache, and uses this information to determine the respective sizes of both parts of the cache.

Growing community of inventors

Barcelona, Spain

Aristides Gionis