Abstract: User behavior modeling can include determining actions performed by various users. From the mined user actions, future actions can be predicted. Certain implementations include providing information and/or services based on the predicted future actions. Some implementations, include providing relevant information, services, and/or goods regarding the predicted future action.

Abstract: A smart user-centric information aggregation system allows a user to define a region of content displayed in a display of a device and performs information aggregation on behalf of the user. The smart user-centric information aggregation system searches, aggregates and groups information related to content included in the region of content for the user while the user can continue to perform his/her original course of actions without interruption. After finding information related to the desired content, the smart user-centric information aggregation system may notify the user and present the found information to the user upon receiving confirmation from the user. The smart user-centric information aggregation system may continue to find new related information and update the presentation with the newly found information periodically, in some instances without user intervention or input.

Abstract: A free-standing zeolite membrane and a zeolite membrane supported on a support structure are disclosed. The free-standing zeolite membrane is fabricated by mixing zeolite particles and an optional inorganic binder, forming a green body, and sintering the green body at a sufficiently low temperature so as to prevent damage to the gas selectivity properties of the zeolite particles. The supported composite zeolite membrane is fabricated by mixing a sacrificial binder, an optional inorganic binder, boehmite sol and zeolite particles to form a slurry. The slurry is then coated onto a porous support structure, dried and sintered at a sufficiently low temperature so as to prevent damage to the gas selective properties of the zeolite particles. In both membranes, the zeolite particles span the entire thickness of the membrane to provide a high selectivity path for the flow of gas to pass therethrough.

Abstract: Techniques for training a non-linear support vector machine utilizing a stochastic gradient descent algorithm are provided. The computations of the stochastic gradient descent algorithm are parallelized via a number of processors. Calculations of the stochastic gradient descent algorithm on a particular processor may be combined according to a packing strategy before communicating the results of the calculations with the other processors.

Abstract: A “General Click Model” (GCM) is constructed using a Bayesian network that is inherently capable of modeling “tail queries” by building the model on multiple attribute values that are shared across queries. More specifically, the GCM learns and predicts user click behavior towards URLs displayed on a query results page returned by a search engine. Unlike conventional click modeling approaches that learn models based on individual queries, the GCM learns click models from multiple attributes, with the influence of different attribute values being measured by Bayesian inference. This provides an advantage in learning that enables the GCM to achieve improved generalization and results, especially for tail queries, than conventional click models. In addition, most conventional click models consider only position and the identity of URLs when learning the model. In contrast, the GCM considers more session-specific attributes in making a final prediction for anticipated or expected user click behaviors.

Abstract: Methods and technologies providing translations of web queries based on an analysis of user behavior in click-through data. These methods and technologies generates large-scale and timely query translation pairs guided by a small set of seed word pairs from a dictionary, without relying on additional knowledge or complex models.

Abstract: A method of manufacturing an edible product, a laminated dough composition and an edible product are provided. The method includes mixing ingredients including starch, gluten, a plasticizer, fat and leavening into a pastry dough composition. A flavored fat composition is prepared by mixing ingredients comprising fat, at least one flavor additive and at least one filling. The flavored fat composition is then applied to a surface of the pastry dough composition. Both compositions are folded and sheeted to form a laminated dough. The laminated dough is then formed into pieces, which are proofed and then baked. A laminated dough composition, and an edible product itself, is provided that comprises a plurality of layers of a flavored fat interposed between a plurality of layers of a pastry dough, the pastry dough composition comprising starch, gluten, a plasticizer, fat and leavening; the flavored fat composition comprising fat, at least one flavor additive and at least one filling.

Abstract: A stable trans fat free emulsifier composition is provided that may have other functional ingredients incorporated to form a base for a flour-based dough preparation to hold the ingredients together. A glyceride emulsifier is heated above its melting point and blended with a non-hydrogenated vegetable oil, or a quantity of glyceride emulsifier is blended with non-hydrogenated vegetable oil and the mixture then heated to a temperature above its melting point. The ratio of emulsifier to vegetable oil is selected to cause the composition, upon cooling, to form a solid plastic mass. The hardness value of the plastic mass is within a range especially applicable to bakery dough applications.

Abstract: The techniques discussed herein consider a degree of noise associated with user clicks performed during search sessions. The techniques then generate a model that characterizes click noise so that search engines can more accurately infer document relevance.

Abstract: User behavior modeling can include determining temporal- or time-based actions performed by various users. From the mined temporal-based user actions, future actions can be predicted. Certain implementations include providing information and/or services based on the predicted future actions. Some implementations, include providing relevant information, services, and/or goods regarding the predicted future action.

Abstract: Described is a technology in which new words (including a phrase or set of Chinese characters) are mined from a query log. The new words may be added to (or otherwise supplement) an IME dictionary. A set of candidate queries may be selected from the log based upon market (e.g., the Chinese market) and/or by language. From this set, various filtering steps are performed to locate only new words that are frequently in used. For example, only frequent queries are kept for further processing, which may include filtering out queries based on length (e.g., less than two or greater than eight Chinese characters), and/or filtering out queries based on too many stop-words in the query. Processing may also include filtering out a query that is a substring of a larger query, or vice-versa. Also described is Pinyin-based clustering and filtering, and filtering out queries already handled in the dictionary.

Abstract: A method of manufacturing an edible product, a laminated dough composition and an edible product are provided. The method includes mixing ingredients including starch, gluten, a plasticizer, fat and leavening into a pastry dough composition. A flavored fat composition is prepared by mixing ingredients comprising fat and at least one flavor additive. The flavored fat composition is then applied to a surface of the pastry dough composition. Both compositions are folded and sheeted to form a laminated dough. The laminated dough is then formed into pieces, which are proofed and then baked. A laminated dough composition, and an edible product itself, is provided that comprises a plurality of layers of a flavored fat interposed between a plurality of layers of a pastry dough, the pastry dough composition comprising starch, gluten, a plasticizer, fat and leavening; the flavored fat composition comprising fat and at least one flavor additive.

Abstract: A method of generating training data for a search engine begins by retrieving log data pertaining to user click behavior. The log data is analyzed based on a click model that includes a parameter pertaining to a user intent bias representing the intent of a user in performing a search in order to determine a relevance of each of a plurality of pages to a query. The relevance of the pages is then converted into training data.

Abstract: Data from a click log may be used to generate training data for a search engine. User click behavior and user post-click behavior may be used to assess the relevance of a page to a query. Labels for training data may be generated based on data from the click log. The labels may pertain to the relevance of a page to a query. For example, user post-click behavior that may be examined includes the amount of time that a user remains on a target page when a user clicks one of the search results.

Abstract: A free-standing zeolite membrane and a zeolite membrane supported on a support structure are disclosed. The free-standing zeolite membrane is fabricated by mixing zeolite particles and an optional inorganic binder, forming a green body, and sintering the green body at a sufficiently low temperature so as to prevent damage to the gas selectivity properties of the zeolite particles. The supported composite zeolite membrane is fabricated by mixing a sacrificial binder, an optional inorganic binder, boehmite sol and zeolite particles to form a slurry. The slurry is then coated onto a porous support structure, dried and sintered at a sufficiently low temperature so as to prevent damage to the gas selective properties of the zeolite particles. In both membranes, the zeolite particles span the entire thickness of the membrane to provide a high selectivity path for the flow of gas to pass therethrough.

Abstract: User behavior modeling can include determining temporal- or time-based actions performed by various users. From the mined temporal-based user actions, future actions can be predicted. Certain implementations include providing information and/or services based on the predicted future actions. Some implementations, include providing relevant information, services, and/or goods regarding the predicted future action.

Abstract: The present invention relates a frozen non-leavened farinaceous dough that can be leavened under ambient conditions before being baked, fried and/or boiled to produce a ready-to-eat foodstuff. More particularly, the invention provides a frozen non-leavened farinaceous dough containing at 103 cells/g viable cells of an S. bayanus strain. The dough product according to the invention offers the advantage that the time window within which the leavened dough can be prepared for consumption is very broad, i.e. several hours. Another aspect of the present invention relates to a method of preparing a ready-to-eat cooked leavened dough product, said method comprising the steps of: ?thawing and leavening a frozen non-leavened dough as described above by keeping it under ambient temperature for 6-24 hours to obtain a leavened dough having a specific volume of at least 2.0 ml/g; and ?baking, frying and/or boiling the leavened dough.

Abstract: Methods, systems, and media are provided for a dynamic batch strategy utilized in parallelization of online learning algorithms. The dynamic batch strategy provides a merge function on the basis of a threshold level difference between the original model state and an updated model state, rather than according to a constant or pre-determined batch size. The merging includes reading a batch of incoming streaming data, retrieving any missing model beliefs from partner processors, and training on the batch of incoming streaming data. The steps of reading, retrieving, and training are repeated until the measured difference in states exceeds a set threshold level. The measured differences which exceed the threshold level are merged for each of the plurality of processors according to attributes. The merged differences which exceed the threshold level are combined with the original partial model states to obtain an updated global model state.

Abstract: A “General Click Model” (GCM) is constructed using a Bayesian network that is inherently capable of modeling “tail queries” by building the model on multiple attribute values that are shared across queries. More specifically, the GCM learns and predicts user click behavior towards URLs displayed on a query results page returned by a search engine. Unlike conventional click modeling approaches that learn models based on individual queries, the GCM learns click models from multiple attributes, with the influence of different attribute values being measured by Bayesian inference. This provides an advantage in learning that enables the GCM to achieve improved generalization and results, especially for tail queries, than conventional click models. In addition, most conventional click models consider only position and the identity of URLs when learning the model. In contrast, the GCM considers more session-specific attributes in making a final prediction for anticipated or expected user click behaviors.

Abstract: Techniques for training a non-linear support vector machine utilizing a stochastic gradient descent algorithm are provided. The computations of the stochastic gradient descent algorithm are parallelized via a number of processors. Calculations of the stochastic gradient descent algorithm on a particular processor may be combined according to a packing strategy before communicating the results of the calculations with the other processors.