Abstract: Sub-queries for a query are determined. The query is for retrieving a data item of a data graph. The data graph stores representations of the data item. Each representation of the data item stores knowledge represented by the data item in a different way or manner. Each sub-query corresponds to a different representation by which the data graph stores the data item. The sub-queries are evaluated to determine an appropriate representation of the data item in fulfillment of the query without duplicatively traversing the data graph, such as by reusing evaluation results of the sub-queries that overlap one another.

Abstract: A table structure corresponding to the table is selected to describe when cells in a cell-pair in the table structure should be similar to one another. A cell similarity function is selected to compare the cells in the cell-pair and output a probability that the cell-pair includes cells that contain values that are similar to one another according to a criteria. A cell similarity probability is determined of a first cell and a second cell in a first cell-pair in the table structure being similar to each other according to the cell similarity function. A computed probability is adjusted of the table structure representing the table using the cell similarity probability. The computed probability is indicated as a probability of a first model representing the table, wherein the first model is a function of the table structure and the selected cell similarity function.

Abstract: A method is provided for analyzing log message content. The computer-implemented method may include detecting, by at least one processor, log messages in an application code, identifying a log level assigned to each of the log messages, and performing natural language processing (NLP) analysis on each of the log messages by using at least keyword and synonym matching percentage analysis criteria. The computer-implemented method may further include determining, in response to the NLP analysis, a severity score of each of the log messages, and reclassifying, based on the severity score, the assigned log level of one or more of the log messages to a different log level.

Abstract: A table structure corresponding to the table is selected to describe when cells in a cell-pair in the table structure should be similar to one another. A cell similarity function is selected to compare the cells in the cell-pair and output a probability that the cell-pair includes cells that contain values that are similar to one another according to a criteria. A cell similarity probability is determined of a first cell and a second cell in a first cell-pair in the table structure being similar to each other according to the cell similarity function. A computed probability is adjusted of the table structure representing the table using the cell similarity probability. The computed probability is indicated as a probability of a first model representing the table, wherein the first model is a function of the table structure and the selected cell similarity function.

Abstract: A semantic search engine is enhanced to employ user preferences to customize answer output by, for a first user, extracting user preferences and sentiment levels associated with a first question; receiving candidate answer results of a semantic search of the first question; weighting the candidate answer results according to the sentiment levels for each of the user preferences; and producing the selected candidate answers to the first user. Optionally, user preferences and sentiment levels may be accumulated over different questions for the same user, or over different users for similar questions. And, supplemental information may be retrieved relative to a user preference in order to further tune the weighting per the preferences and sentiment levels.

Abstract: A semantic search engine is enhanced to employ user preferences to customize answer output by, for a first user, extracting user preferences and sentiment levels associated with a first question; receiving candidate answer results of a semantic search of the first question; weighting the candidate answer results according to the sentiment levels for each of the user preferences; and producing the selected candidate answers to the first user. Optionally, user preferences and sentiment levels may be accumulated over different questions for the same user, or over different users for similar questions. And, supplemental information may be retrieved relative to a user preference in order to further tune the weighting per the preferences and sentiment levels.

Abstract: Methods, apparatus and systems, including computer program products, for creating subject matter synonyms from definitions extracted from a subject matter glossary. Confidence scores, each representing a likelihood that two terms defined in the subject matter glossary are synonyms, are determined by applying natural language processing (e.g., passage term matching, lexical matching, and syntactic matching) to the extracted definitions. A subject matter thesaurus is built based on the confidence scores. In one embodiment, a statement containing a first term is created based on an extracted definition of the first term, a modified statement is created by substituting a second term in the statement in lieu of the first term, a corpus is searched, and a confidence score is determined based on evidence in the corpus that the modified statement is accurate. The first and second terms are marked as synonyms if the confidence score is greater than a threshold.

Abstract: A system, and computer program product for subject-matter analysis of tabular data are provided in the illustrative embodiments. A first document including the tabular data is received. A library of functional signatures for a first subject-matter domain is selected. A determination is made whether a threshold number of functional signatures from the selected library are applicable to the tabular data, wherein a functional signature is applicable to the tabular data when values in the tabular data correspond to an operation and a table structure specified in the functional signature. Responsive to the threshold number of functional signatures from the selected library being applicable to the tabular data, a processor and a memory process the first document according to a process for the first subject matter domain selected from a plurality of processes for respective subject matter domains.

Abstract: A method for discovering relationships in tabular data is provided in the illustrative embodiments. A set of documents is received, a document in the set including the tabular data. A cell in the tabular data is selected whose dependencies are to be determined. A hypothesis to use in conjunction with the cell is selected. Whether the hypothesis applies to a selected portion of the document is tested by determining whether a conclusion in the hypothesis can be computed using a function specified in the hypothesis on the selected portion. The selected portion can be a selected cell-range in the tabular data or content in a non-tabular portion of the document. The hypothesis is utilized to describe the cell relative to the selected portion.

Abstract: A system, and computer program product for discovering relationships in tabular data are provided in the illustrative embodiments. A set of documents is received, a document in the set including the tabular data. A cell in the tabular data is selected whose dependencies are to be determined. A hypothesis to use in conjunction with the cell is selected. Whether the hypothesis applies to a selected portion of the document is tested by determining whether a conclusion in the hypothesis can be computed using a function specified in the hypothesis on the selected portion. The selected portion can be a selected cell-range in the tabular data or content in a non-tabular portion of the document. The hypothesis is utilized to describe the cell relative to the selected portion.

Abstract: A method, system, and computer program product for subject-matter analysis of tabular data are provided in the illustrative embodiments. A first document including the tabular data is received. A library of functional signatures for a first subject-matter domain is selected. A determination is made whether a threshold number of functional signatures from the selected library are applicable to the tabular data, wherein a functional signature is applicable to the tabular data when values in the tabular data correspond to an operation and a table structure specified in the functional signature. Responsive to the threshold number of functional signatures from the selected library being applicable to the tabular data, a processor and a memory process the first document according to a process for the first subject matter domain selected from a plurality of processes for respective subject matter domains.

Abstract: Visualizable subject matter contained within data is detected. A context in which the visualizable subject matter appears is determined. A visualization engine of a plurality of visualization engines is selected based, at least in part, on the context in which the visualizable subject matter appears. The visualization engine is configured to generate a visualization of the visualizable subject matter.

Abstract: A mechanism is provided in a data processing system for aggregating sentiment about an entity from a corpus of documents. The mechanism identifies a plurality of sentiment passages in the corpus of documents. Each of the plurality of sentiment passages includes a statement of sentiment about the entity. The mechanism determines a plurality of passage sentiment scores for the plurality of sentiment passages and an actual aggregate sentiment score from the plurality of passage sentiment scores based on a k-valued model. The mechanism determines a sentiment confidence score for the actual aggregate sentiment score based on the raw aggregate sentiment score and the actual aggregate sentiment score and presents the actual aggregate sentiment score and the sentiment confidence score.

Abstract: Mechanisms are provided for performing tabular data correction in a document. Tabular data is received and analyzed to identify at least one portion of the tabular data having an erroneous/missing data value. A functional dependency of the at least one portion of the tabular data on one or more other portions of the tabular data is determined. A correct data value for the erroneous or missing data value of the at least one portion of the tabular data is determined based on the functional dependency of the at least one portion. In addition, the tabular data is modified to replace the erroneous or missing data value with the correct data value and thereby generate a modified table data. A processing operation is then performed on the modified table data to generate a resulting output.

Abstract: A semantic search engine is enhanced to employ user preferences to customize answer output by, for a first user, extracting user preferences and sentiment levels associated with a first question; receiving candidate answer results of a semantic search of the first question; weighting the candidate answer results according to the sentiment levels for each of the user preferences; and producing the selected candidate answers to the first user. Optionally, user preferences and sentiment levels may be accumulated over different questions for the same user, or over different users for similar questions. And, supplemental information may be retrieved relative to a user preference in order to further tune the weighting per the preferences and sentiment levels.

Abstract: A semantic search engine is enhanced to employ user preferences to customize answer output by, for a first user, extracting user preferences and sentiment levels associated with a first question; receiving candidate answer results of a semantic search of the first question; weighting the candidate answer results according to the sentiment levels for each of the user preferences; and producing the selected candidate answers to the first user. Optionally, user preferences and sentiment levels may be accumulated over different questions for the same user, or over different users for similar questions. And, supplemental information may be retrieved relative to a user preference in order to further tune the weighting per the preferences and sentiment levels.

Abstract: An approach is provided in which a knowledge manager trains a machine-learning model and generates a hyperplane based upon a first set of labeled feature vectors. The knowledge manager computes, relative to the hyperplane, a first distribution of a first set of feature vectors corresponding to a first set of source documents. Subsequently, the knowledge manager computes, relative to the hyperplane, a second distribution of a second set of feature vectors corresponding to a second group of source documents. The knowledge manager, in turn, generates an indicator to retrain the machine-learning model in response to determining that a distribution difference between the second distribution and the first distribution reaches a distribution difference threshold.

Abstract: Methods, apparatus and systems, including computer program products, for creating subject matter synonyms from definitions extracted from a subject matter glossary. Confidence scores, each representing a likelihood that two terms defined in the subject matter glossary are synonyms, are determined by applying natural language processing (e.g., passage term matching, lexical matching, and syntactic matching) to the extracted definitions. A subject matter thesaurus is built based on the confidence scores. In one embodiment, a statement containing a first term is created based on an extracted definition of the first term, a modified statement is created by substituting a second term in the statement in lieu of the first term, a corpus is searched, and a confidence score is determined based on evidence in the corpus that the modified statement is accurate. The first and second terms are marked as synonyms if the confidence score is greater than a threshold.

Abstract: An approach is provided to identify permanence data corresponding to terms included in a corpus of a question answering (QA) system. Based on the identified permanence, a time-based confidence of each of the terms is established. Terms are identified as a plurality of candidate answers to a question posed to the QA system. Each of the plurality of candidate answers are scored with the scoring being at least partially based on the time-based confidence established for each of the terms.

Abstract: An approach is provided to identify permanence data corresponding to terms included in a corpus of a question answering (QA) system. Based on the identified permanence, a time-based confidence of each of the terms is established. Terms are identified as a plurality of candidate answers to a question posed to the QA system. Each of the plurality of candidate answers are scored with the scoring being at least partially based on the time-based confidence established for each of the terms.