Abstract: Techniques for creating an interest graph include obtaining content items from multiple content sources and applying tailored (e.g., source-specific) preprocessing to the content items based on their respective content source. Text is extracted and salient keywords and key phrases are identified using unsupervised machine learning models. The keywords and key phrases become nodes in an interest graph, each node comprising an embedding of a keyword or key phrase in a common embedding space, with edges representing semantic similarity based on embeddings or co-engagement patterns. The graph provides an expansive, granular, and dynamic taxonomy easily adaptable to emerging interests. The interest graph overcomes limitations of conventional taxonomies that lack depth, fail to capture niche interests, and cannot adapt to reflect evolving user preferences. The described techniques construct a rich interest graph from diverse content for improved content understanding.

Abstract: Techniques for automated tagging of visual content are described. A pairwise model is used to encode images and videos with a vision encoder, and encode keywords and phrases from an interest graph with a text encoder. Similarity layers compare these cross-modality embeddings by calculating distance in a shared embedding space. Scores indicate the association between visual features and text. A pairwise loss function brings together matched pairs while separating non-matches during training. Scores exceeding a threshold tag content with relevant keywords and phrases. The pairwise architecture relates images and text despite limited associated text. It leverages vision-to-text understanding for accurate tagging without per-class labels. Contrastive similarity techniques associate visual patterns with textual concepts. Automated tagging organizes user-generated content by topics using this scalable cross-modality approach.

Abstract: A method and system for classifying a triage-related message related to a software application security technical problem is provided. A triage-related classification is generated for the triage-related message by applying a processor-implemented machine learning model that has been trained to analyze the text of the triage-related message. The generated triage-related classification is sent to a user for remediating the software application security technical problem.

Abstract: The present invention relates to an adsorption moisture pump based air to water harvesting device and a method of harvesting water from ambient air. The water harvesting device [1100] comprises a rotary desiccant unit, a heat pump unit [1104] and a control unit. The rotary desiccant unit comprises a desiccant wheel [102], a reactivation air inlet [1108a], a reactivation air outlet [1108b], a process air inlet [1106a] and a process air outlet [1106b]. The desiccant wheel [1102] comprises at least a process sector [1106] and a reactivation sector [1108] and a wheel drive. The heat pump unit [1104] comprises at least one compressor, an expansion valve [1116], an evaporator [1112], a main condenser [1110], and such that a refrigerant fluid is flown sequentially within the compressor [1114], the main condenser [1110], the expansion valve [1116], and the evaporator [1112].

Abstract: Described herein is an apparatus and method for removing moisture and/or sorbates from airstream and/or other fluid. The system comprises a preconditioning desiccant wheel and a main desiccant wheel. The preconditioning desiccant wheel comprises of at least two first sectors for allowing air to pass therethrough, including a first process sector and a first reactivation sector. The main desiccant wheel comprises of at least three second sectors for allowing air to pass therethrough, sequentially including a second outside air sector, a second process sector, and a second reactivation sector. The ambient airstream is sequentially dehumidified in one of the at least two sectors of the preconditioning desiccant wheel and the outside air sector of the main desiccant wheel, before being at least partially supplied for reactivation to the reactivation sector of the main desiccant wheel.

Abstract: Described herein are systems, methods, and software to manage the compression of route tables for communication between networking elements. In one implementation, a network device identifies network keys for a route table by replacing attributes in the tables with values. The network device further generates a compressed route table using the route keys and associating each of the route keys with one or more additional attributes. The network device also generates a dictionary to associate each of the values for the route keys to a corresponding attribute of the attributes.

Abstract: Described herein are systems, methods, and software to manage the compression of route tables for communication between networking elements. In one implementation, a network device identifies network keys for a route table by replacing attributes in the tables with values. The network device further generates a compressed route table using the route keys and associating each of the route keys with one or more additional attributes. The network device also generates a dictionary to associate each of the values for the route keys to a corresponding attribute of the attributes.

Abstract: Systems and techniques for sequence to sequence conversational query understanding are described herein. A query may be received that includes multiple words. It may be identified that the query is to be reformulated based on an attention value for an attention word in the query. Relationships may be determined among words of the query and words in a previously submitted query and words in results from the previously submitted query. The query may be reformulated based on the relationships. The reformulated query may be employed to retrieve query results.

Abstract: Representative embodiments disclose mechanisms to provide direct answers to a query submitted by a user. The mechanisms are tailored so that the answers presented have a high confidence of being correct. A plurality of document segments that are relevant to the query are selected. The selected segments are submitted to a trained machine reading comprehension model along with the query. The result is an extracted answer for one or more of the submitted segments. A subset of the extracted answers are clustered and an answer for each cluster having at least a threshold number of answers are selected as direct answers. The direct answers are presented in a format suitable to the number of selected direct answers.

Abstract: Examples of the present disclosure describe systems and methods relating to generating a relevance score on a given natural language answer to a natural language query for ranking the answer among other answers for the query, while generating a summary passage and a likely query to the given passage. For instance, multi-layered, recurrent neural networks may be used to encode the query and the passage, along with a multi-layered neural network for information retrieval features, to generate a relevant score for the passage. A multi-layered, recurrent neural network with soft attention and sequence-to-sequence learning task may be used as a decoder to generate a summary passage. A common encoding neural network may be employed to encode the passage for the ranking and the summarizing, in order to present concise and accurate natural language answers to the query.

Abstract: Non-limiting examples of the present disclosure describe a unified ranking model that may be used by a plurality of entry points to return ranked results in response to received query data. The unified ranking model is provided as a service for a plurality of entry points. A query is received from an entry point of the plurality of entry points. Results data for the query data is retrieved. A unified ranking model is executed to rank the results data. Execution of the unified ranking model manipulates feature data of the unified ranking model based on user context signals associated with the received query data and acquired result retrieval signals corresponding with the retrieved results data. Execution of the unified ranking model generates ranked result data. Ranked results data is returned to the processing device corresponding with the entry point. Other examples are also described.

Abstract: Representative embodiments disclose mechanisms to provide direct answers to a query submitted by a user. The mechanisms are tailored so that the answers presented have a high confidence of being correct. A plurality of document segments that are relevant to the query are selected. The selected segments are submitted to a trained machine reading comprehension model along with the query. The result is an extracted answer for one or more of the submitted segments. A subset of the extracted answers are clustered and an answer for each cluster having at least a threshold number of answers are selected as direct answers. The direct answers are presented in a format suitable to the number of selected direct answers.

Abstract: Systems and techniques for sequence to sequence conversational query understanding are described herein. A query may be received that includes multiple words. It may be identified that the query is to be reformulated based on an attention value for an attention word in the query. Relationships may be determined among words of the query and words in a previously submitted query and words in results from the previously submitted query. The query may be reformulated based on the relationships. The reformulated query may be employed to retrieve query results.

Abstract: Examples of the present disclosure describe systems and methods relating to generating a relevance score on a given natural language answer to a natural language query for ranking the answer among other answers for the query, while generating a summary passage and a likely query to the given passage. For instance, multi-layered, recurrent neural networks may be used to encode the query and the passage, along with a multi-layered neural network for information retrieval features, to generate a relevant score for the passage. A multi-layered, recurrent neural network with soft attention and sequence-to-sequence learning task may be used as a decoder to generate a summary passage. A common encoding neural network may be employed to encode the passage for the ranking and the summarizing, in order to present concise and accurate natural language answers to the query.

Abstract: Examples of the present disclosure describe systems and methods relating to generating a relevance score on a given natural language answer to a natural language query for ranking the answer among other answers for the query, while generating a summary passage and a likely query to the given passage. For instance, multi-layered, recurrent neural networks may be used to encode the query and the passage, along with a multi-layered neural network for information retrieval features, to generate a relevant score for the passage. A multi-layered, recurrent neural network with soft attention and sequence-to-sequence learning task may be used as a decoder to generate a summary passage. A common encoding neural network may be employed to encode the passage for the ranking and the summarizing, in order to present concise and accurate natural language answers to the query.

Abstract: Conversational or multi-turn question understanding using web intelligence is provided. An intelligent query understanding system is provided for receiving a context-dependent query from a user, obtaining contextual information related to the context-dependent query, and reformatting the context-dependent query as one or more reformulations based on the contextual information. The intelligent query understanding system is further operative to query a search engine with the one or more reformulations, receive one or more candidate results, and select a highest ranked reformulation based on the candidate results. The system can provide the highest ranked reformulation of the highest ranked reformulation as a response.

Abstract: Examples of the present disclosure describe systems and methods relating to generating a relevance score on a given natural language answer to a natural language query for ranking the answer among other answers for the query, while generating a summary passage and a likely query to the given passage. For instance, multi-layered, recurrent neural networks may be used to encode the query and the passage, along with a multi-layered neural network for information retrieval features, to generate a relevant score for the passage. A multi-layered, recurrent neural network with soft attention and sequence-to-sequence learning task may be used as a decoder to generate a summary passage. A common encoding neural network may be employed to encode the passage for the ranking and the summarizing, in order to present concise and accurate natural language answers to the query.

Abstract: Aspects of the technology described herein increase the efficiency of a search session by determining whether fresh content is likely to be responsive to the user's query. Whether fresh content is likely to be responsive to a specific query is determined by retrieving social media posts that are responsive to the query. The social media posts are evaluated for virality, which is the tendency of a social media post to be circulated rapidly and widely from one Internet user to another. The virality of a social media post can be determined by comparing a number of times the social media post has been re-communicated by individual users. Queries that return viral social media posts may be classified as seeking fresh content.

Abstract: Non-limiting examples of the present disclosure describe a unified ranking model that may be used by a plurality of entry points to return ranked results in response to received query data. The unified ranking model is provided as a service for a plurality of entry points. A query is received from an entry point of the plurality of entry points. Results data for the query data is retrieved. A unified ranking model is executed to rank the results data. Execution of the unified ranking model manipulates feature data of the unified ranking model based on user context signals associated with the received query data and acquired result retrieval signals corresponding with the retrieved results data. Execution of the unified ranking model generates ranked result data. Ranked results data is returned to the processing device corresponding with the entry point. Other examples are also described.

Abstract: A natural language query may be transformed to a transformed natural language while keeping sufficient semantic meaning such that the query may be transformed. A natural language query may be received by a computing device and sent to a natural language transformation model for transformation. The transformation may use a variety of techniques including stop word removal, stop structure removal, noun phrase/entity detection, key concept detection, dependency filtering. The techniques may be sequenced.