Abstract: An autonomous vehicle utilizing machine learning method, system, and computer program product include training, by a computing device, a driver attention model utilizing historical ambient sensor data and historical physiological sensor data associated with an autonomous vehicle, receiving ambient sensor data associated with a currently driving autonomous vehicle, receiving physiological sensor data associated with a driver of the currently driving autonomous vehicle, and using, by a computing device, the trained driver attention model in conjunction with the ambient sensor data and the physiological sensor data to generate an alert for a driver of the currently driving autonomous vehicle.

Abstract: Systems and techniques that facilitate spurious relationship filtration from external knowledge graphs based on distributional semantics of an input corpus are provided. In one or more embodiments, a context component can generate a context-based word embedding of one or more first terms in a document collection. The embedding can yield vector representations of the one or more first terms. The one or more first terms can correspond to knowledge terms in one or more first nodes of a knowledge graph. In one or more embodiments, a filtering component can filter out a relationship between the one or more first nodes and a second node of the knowledge graph based on a similarity value being less than a threshold. The similarity value can be a function of the vector representations of the one or more first terms. In various embodiments, cosine similarity can be used to compute the similarity value.

Abstract: A computer-implemented method according to one embodiment includes converting an input question into a vector form using trained word embeddings; constructing a type similarity matrix using a predetermined ontology; and determining a score for all possible types for the input question, based on the input question in vector form and the type similarity matrix.

Abstract: An approach is provided for signaling concept drift during knowledge base population. A knowledge graph and a collection of text is received, and a vector space is built. A sequence of data items associated with a type of entity or a relation is received. Entities or relations from the knowledge graph are embedded into the vector space to generate entity or relation vectors. Data items associated with the type of entity or the relation are embedded into the vector space to generate data item vectors. An emerging entity or relation concept vector is computed by determining a centroid of the data item vectors. An entity or relation concept vector is computed by determining a centroid of the entity or relation vectors. A signal is generated when a distance between the emerging entity or relation concept vector and the entity or relation concept vector is greater than a threshold.

Abstract: A method, a computer program product, and a computer system determine when to transfer a communication session from an agent to a bot. The method includes monitoring the communication session between the agent and a user. The method includes determining a continuing utility value indicating a predicted continuing cost to maintaining the communication session with the agent. The continuing utility value is indicative of a predicted continuing benefit to maintaining the communication with the agent. The method includes determining a transferring utility value indicating a predicted transferring cost to transferring the communication session from the agent to the bot. The transferring utility value is indicative of a predicted transferring benefit to transferring the communication session from the agent to the bot. The method includes, as a result of the predicted transferring benefit being greater than the predicted continuing benefit, transferring the communication session from the agent to the bot.

Abstract: A query-focused faceted structure generation method, system, and computer program product for generating a query-focused faceted structure from a taxonomy for searching a document corpus, including augmenting taxonomy types with new instances where the instances comprise entities within a proximity of existing instances of taxonomy types in a local embedding of entities parsed from the document corpus, ranking each instance in the augmented taxonomy with respect to its type as a function of both a distance from an instance to a query in a global embedding vector space of the entities trained from the document corpus and a distance of an instance to a type in the local embedding, and ranking the taxonomy types using expanded instances in the document corpus for each type.

Abstract: Training a machine learning model such as a neural network, which can automatically extract a hypernym from unstructured data, is disclosed. A preliminary candidate list of hyponym-hypernym pairs can be parsed from the corpus. A preliminary super-term-sub-term glossary can be generated from the corpus, the preliminary super-term-sub-term glossary containing one or more super-term-sub-term pairs. A super-term-sub-term pair can be filtered from the preliminary super-term-sub-term glossary, responsive to detecting that the super-term-sub-term pair is not a candidate for hyponym-hypernym pair, to generate a final super-term-sub-term glossary. The preliminary candidate list of hyponym-hypernym pairs and the final super-term-sub-term glossary can be combined to generate a final list of hyponym-hypernym pairs. An artificial neural network can be trained using the final list of hyponym-hypernym pairs as a training data set, the artificial neural network trained to identify a hypernym given new text data.

Abstract: Systems and techniques that facilitate spurious relationship filtration from external knowledge graphs based on distributional semantics of an input corpus are provided. In one or more embodiments, a context component can generate a context-based word embedding of one or more first terms in a document collection. The embedding can yield vector representations of the one or more first terms. The one or more first terms can correspond to knowledge terms in one or more first nodes of a knowledge graph. In one or more embodiments, a filtering component can filter out a relationship between the one or more first nodes and a second node of the knowledge graph based on a similarity value being less than a threshold. The similarity value can be a function of the vector representations of the one or more first terms. In various embodiments, cosine similarity can be used to compute the similarity value.

Abstract: In one embodiment, in accordance with the present invention, a method, computer program product, and system for performing actions based on captured interpersonal interactions during a meeting is provided. One or more computer processors capture the interpersonal interactions between people in a physical space during a period of time, using machine learning algorithms to detect the interpersonal interactions and a state of each person based on vision and audio sensors in the physical space. The one or more computer processors analyze and categorize the interactions and state of each person, and tag representations of each person with the respectively analyzed and categorized interactions and states of the respective person over the period of time. The one or more computer processors then take an action based on the analysis.

Abstract: Mechanisms are provided for conducting a natural language dialogue between the automatic dialogue system and a user of a client computing device. An automatic dialogue system receives natural language text corresponding to a user input from the user via the client computing device, the natural language text having an ambiguous portion of natural language text. The automatic dialogue system analyzes user profile information corresponding to the user to identify an anomaly in the user profile information and predicts a user intent associated with the anomaly. The automatic dialogue system disambiguates the ambiguous portion of the natural language text based on the predicted user intent and generates a response to the user input based on the disambiguated natural language text which is output to the client computing device to thereby conduct the natural language dialogue.

Abstract: Mechanisms are provided for customizing responses to future questions based on identified anomalies in user profile information. An automated dialogue system monitors information associated with a plurality of entities, where the information includes quantities for variable values associated with the entities. The automated dialogue system, in response to determining that a quantity of a variable value associated with an entity in the plurality of entities has changed by an amount equal to or exceeding a corresponding threshold value, generates response information associated with a quantity of the variable value and an entity to respond to at least one future question. In addition, the automated dialogue system stores the responsive information in association with the entity for later retrieval in response to initiation of a dialogue session with the automated dialogue system. Time thresholds may be established for determining when to stop using the responsive information for responding to questions.

Abstract: A method, apparatus and computer program product for responding to an indirect utterance in a dialogue between a user and a conversational system is described. An indirect utterance is received. A parse structure of the indirect utterance is generated. The indirect utterance is an utterance which does not match a user goal expressed as elements of a knowledge graph. The parse structure is connected through the knowledge graph to a user goal to issue a user request which is not stated in the indirect utterance. The parse structure is connected using a matching process which matches the parse structure with the connected user goal in the knowledge graph according to a similarity of the parse structure and a portion of the knowledge graph including the connected user goal. A system response is performed based on the connected user goal.

Abstract: A query-focused faceted structure generation method, system, and computer program product for generating a query-focused faceted structure from a taxonomy for searching a document collection, including ingesting a document corpus, generating a vector space representation of a query and instances from a taxonomy of the document corpus, and producing a dynamic structure of a relevant facet categories and facet values using a two-vector space representation from the generated vector space representation.

Abstract: A query-focused faceted structure generation method, system, and computer program product for generating a query-focused faceted structure from a taxonomy for searching a document corpus, including augmenting taxonomy types with new instances where the instances comprise entities within a proximity of existing instances of taxonomy types in a local embedding of entities parsed from the document corpus, ranking each instance in the augmented taxonomy with respect to its type as a function of both a distance from an instance to a query in a global embedding vector space of the entities trained from the document corpus and a distance of an instance to a type in the local embedding, and ranking the taxonomy types using expanded instances in the document corpus for each type.

Abstract: Methods, systems, and computer program products for expressive temporal predictions over semantically-driven time windows are provided herein. A computer-implemented method includes identifying, within a knowledge graph pertaining to a given prediction, a subset of the knowledge graph related to one or more predicted training examples, wherein the subset comprises (i) a set of nodes and (ii) one or more relationships among the set of nodes; determining, for the identified subset, one or more snapshots of the knowledge graph relevant to the given prediction; quantifying a validity window for the one or more predicted training examples, wherein the validity window comprises a temporal bound for prediction validity; and computing a validity window for the given prediction based on the quantified validity window for the one or more predicted training examples.

Abstract: Mechanisms are provided for customizing responses to future questions based on identified anomalies in user profile information. An automated dialogue system monitors information associated with a plurality of entities, where the information includes quantities for variable values associated with the entities. The automated dialogue system, in response to determining that a quantity of a variable value associated with an entity in the plurality of entities has changed by an amount equal to or exceeding a corresponding threshold value, generates response information associated with a quantity of the variable value and an entity to respond to at least one future question. In addition, the automated dialogue system stores the responsive information in association with the entity for later retrieval in response to initiation of a dialogue session with the automated dialogue system. Time thresholds may be established for determining when to stop using the responsive information for responding to questions.

Abstract: A spreadsheet population method, system, and computer program product include associating text with a spreadsheet, the text including candidate data items for populating the spreadsheet, building a multi-dimensional analogy model where each dimension comprises a unique pair of data items where the data items co-occur within a same context window, accepting example data items in the spreadsheet where the data items form tuples in a same implicit relationship according to a spatial configuration, and performing an assistance operation on the spreadsheet using the data item tuples retrieved using the analogy model from the example data items.

Abstract: Methods, systems, and computer program products for linkage prediction through similarity analysis are provided herein. A computer-implemented method includes extracting multiple features from (i) one or more attributes of a set of source nodes within a knowledge graph and (ii) one or more attributes of a set of target nodes within the knowledge graph, wherein at least one extracted feature satisfies a designated complexity level; performing a similarity analysis across the at least one extracted feature by applying one or more similarity measures to the at least one extracted feature; predicting one or more sets of links between the source nodes and the target nodes based on the similarity analysis, wherein one or more sets of predicted links satisfy a pre-determined accuracy threshold; and outputting the one or more sets of predicted links to a user.

Abstract: Mechanisms are provided for conducting a natural language dialogue between the automatic dialogue system and a user of a client computing device. An automatic dialogue system receives natural language text corresponding to a user input from the user via the client computing device, the natural language text having an ambiguous portion of natural language text. The automatic dialogue system analyzes user profile information corresponding to the user to identify an anomaly in the user profile information and predicts a user intent associated with the anomaly. The automatic dialogue system disambiguates the ambiguous portion of the natural language text based on the predicted user intent and generates a response to the user input based on the disambiguated natural language text which is output to the client computing device to thereby conduct the natural language dialogue.

Abstract: A first neural network is operated on a processor and a memory to encode a first natural language string into a first sentence encoding including a set of word encodings. Using a word-based attention mechanism with a context vector, a weight value for a word encoding within the first sentence encoding is adjusted to form an adjusted first sentence encoding. Using a sentence-based attention mechanism, a first relationship encoding corresponding to the adjusted first sentence encoding is determined. An absolute difference between the first relationship encoding and a second relationship encoding is computed. Using a multi-layer perceptron, a degree of analogical similarity between the first relationship encoding and a second relationship encoding is determined.