Abstract: Aspects include engaging a user in a chat flow by a dialog system. A user sentiment associated with a system response provided by the dialog system as part of the chat flow is determined based on observation of the user. A next system response is rerouted from a planned sequence of the chat flow to a sentiment-based repair sequence to alter content delivered to the user based on a detected aspect of the user sentiment.

Abstract: Curation of a corpus of a cognitive computing system is performed interactively by reporting on user interface device to a user a parse tree illustration of discrepancies and corresponding assigned confidence factors detected between at least a portion of a first document and a second or more documents in the corpus. Responsive to a user selection of an illustrated discrepancy in the parse tree, a drill-down dialog is prepared and displayed which shows at least a text string for the portion of the first document and at least one conflicting text string from the second or more documents, and which provides at least one user-selectable administrative action option for handling the detected discrepancy. Responsive to receipt of user selection of an administrative action option, the computing system performs the action to handle the detected discrepancy.

Abstract: Aspects include determination of automated response follow-up. A response to a question is received at a response follow-up system. The response follow-up system analyzes the response using natural language processing to identify one or more response terms. The response follow-up system determines one or more follow-up questions based on the one or more response terms. The response follow-up system modifies an aspect of a user interface based on the one or more follow-up questions.

Abstract: Aspects include generation of automated response follow-up. A response to a question is received at a response follow-up system. The response follow-up system analyzes the response using natural language processing to identify one or more response terms. The response follow-up system generates one or more follow-up questions based on the one or more response terms. The response follow-up system modifies an aspect of a user interface based on the one or more follow-up questions.

Abstract: Context data is determined for a web page. The context data includes textual data that describes subject matter displayed on the web page. A client query is retrieved from a client-retrieved instance of the web page. An identifier associated with the web page capable of distinguishing the web page from other web pages is determined. In response to receiving the client query, a plurality of queries are retrieved from a database. Whether one or more of the retrieved queries correlate the determined web page context and the received client query is determined. A set of one or more queries from the plurality of queries is selected based on the selected set of queries each having been determined to correlate the web page context data and the client query. The selected set of the plurality of the queries is sent to the client-retrieved instance of the web page.

Abstract: Aspects include cross validation of user feedback in a dialog system. A repository of previous user feedback associated with a user interface is accessed. One or more identified features in the previous user feedback are classified. A sequence of cross-validation questions is generated in a dialog system to elicit further feedback from a current user of the user interface based on the one or more identified features and observed interactions of the current user with the user interface. Responses to the sequence of cross-validation questions are filtered to group the further feedback associated with the one or more identified features. One or more change suggestions to modify the user interface are stored based on the previous user feedback and the further feedback associated with the one or more identified features.

Abstract: Curation of a corpus of a cognitive computing system is performed interactively by reporting on user interface device to a user a parse tree illustration of discrepancies and corresponding assigned confidence factors detected between at least a portion of a first document and a second or more documents in the corpus. Responsive to a user selection of an illustrated discrepancy in the parse tree, a drill-down dialog is prepared and displayed which shows at least a text string for the portion of the first document and at least one conflicting text string from the second or more documents, and which provides at least one user-selectable administrative action option for handling the detected discrepancy. Responsive to receipt of user selection of an administrative action option, the computing system performs the action to handle the detected discrepancy.

Abstract: A selected document from corpus of a cognitive computing system is processed according to its association with a category of documents already contained within the corpus, such as a topical category. Then, the cognitive computing system is engaged automatically to query one or more questions previously-associated with the category to discover discrepancies between the selected document and the other corpus documents. If a discrepancy is found, a confidence factor based upon extracted natural language entities and relationships is assigned to the discrepancy, and it is flagged and reported to a user for reconciliation.

Abstract: Mechanisms are provided, in a Question and Answer (QA) system, for identifying a related previously processed question. A first question is processed by the QA system to generate an answer to the first question and a supporting evidence passage that supports the answer to the first question being a correct answer to the first question. The QA system performs a lookup operation in a question repository based on an identifier of the supporting evidence passage to identify an entry corresponding to a previously processed question, different than the first question, that previously had an answer generated based on the supporting evidence passage. The QA system returns the previously processed question and the answer to the first question to a submitter of the first question.

Abstract: A new document is processed for ingestion into a corpus of a cognitive computing system by associating with a category of documents already contained within the corpus, such as a topical category. Then, the cognitive computing system is engaged automatically to query one or more questions previously-associated with the category to discover discrepancies between the new document and those already in the corpus. If a discrepancy is found, a confidence factor based upon extracted natural language entities and relationships is assigned to the discrepancy, and it is flagged and reported to a user for reconciliation.

Abstract: A method, system and a computer program product are provided for collecting ground truth data by activating a ground truth collection browser plugin with a designated toolbar button to receive answer data from a webpage which includes user-selected data passage and source data from the webpage for automatic loading into a ground truth submission window along with user input question data corresponding to the answer data so that the user input question data, source data, and answer data from the ground truth submission window may be stored as ground truth data.

Abstract: A method, system and a computer program product are provided for collecting document segmentation data by activating a document segmentation collection browser plugin with a designated toolbar button to generate one or more initial document segments from a webpage document and to receive user feedback for modifying a first initial document segment through a document segment control tool to generate a modified set of one or more initial document segments which are stored as document and document preprocessing data for the webpage document.

Abstract: A technique for instance-specific feature-based cross-document sentiment aggregation includes analyzing input data to determine an entity referenced in the input data. One or more salient features of the entity are identified. Multiple documents that reference the entity are located. The salient features in each of the multiple documents are identified. Respective sentiment values are assigned to each of the salient features identified in the multiple documents. Respective sub-graphs are created for each of the multiple documents. The sub-graphs specify the sentiment values assigned to each of the salient features identified in an associated one of the multiple documents. The assigned sentiment values for each of the salient features are aggregated based on the sub-graphs. Finally, output data that is associated with the input data is generated. The output data provides an indication of the aggregated sentiment value for each of the salient features.

Abstract: A mechanism is provided in a data processing system for automatically generating question and answer pairs for training a question answering system for a given domain. The mechanism receives user input of question text for a question to be submitted to a question answering system in a user interface. The mechanism determines a question strength score for the question text. The question strength score represents a likelihood the question text will result in a correct answer with high confidence. The mechanism presents a graphical representation of the question strength score in the user interface.

Abstract: A mechanism is provided in a data processing system for generating poll information from a chat session. The mechanism monitors chat content in the multiple user chat session. The mechanism identifies a poll question in a message within the chat content. The mechanism identifies a plurality of answers to the poll question in subsequent messages within the chat content. The mechanism generates a poll user interface presenting the poll question and the plurality of answers as selectable options and presents the poll user interface to the multiple users of the chat session.

Abstract: A mechanism is provided in a data processing system for generating poll information from a chat session. The mechanism monitors chat content in the multiple user chat session. The mechanism identifies a poll question in a message within the chat content. The mechanism identifies a plurality of answers to the poll question in subsequent messages within the chat content. The mechanism generates a poll user interface presenting the poll question and the plurality of answers as selectable options and presents the poll user interface to the multiple users of the chat session.

Abstract: Context data is determined for a web page. The context data includes textual data that describes subject matter displayed on the web page. A client query is retrieved from a client-retrieved instance of the web page. An identifier associated with the web page capable of distinguishing the web page from other web pages is determined. In response to receiving the client query, a plurality of queries are retrieved from a database. Whether one or more of the retrieved queries correlate the determined web page context and the received client query is determined. A set of one or more queries from the plurality of queries is selected based on the selected set of queries each having been determined to correlate the web page context data and the client query. The selected set of the plurality of the queries is sent to the client-retrieved instance of the web page.

Abstract: A domain specific ontology collection associated with a user is determined. At least one action of the user is received. A task for the user is determined based on the domain specific ontology collection for the user and the at least one action of the user.

Abstract: An answer to an input question may be formulated using a first corpus of information. Using the answer, a group of candidate images related to the answer from a second corpus of information may be identified. Using the answer and the group of candidate images, a group of modified images may be generated. Generating modified images may include marking, with a visual indicator, a portion of content in at least one image from the group of candidate images.

Abstract: A method for providing content relevant images for an input question to a deep question answering system is disclosed. The method can include formulating, in response to receiving the input question, an answer to the input question. The method can also include identifying, based on the answer to the input question, a set of candidate images from a corpus of information. The set of candidate images may relate to the answer. The method can also include determining, based on a confidence value of the set of candidate images and a designated display specification, a first subset of the set of candidate images to provide in a display area. The method can also include selecting the first subset to be provided in the display area.