This work aims to address the problem of image-based question-answering (QA) with new models and datasets. In our work, we propose to use neural networks and visual semantic embeddings, without intermediate stages such as object detection and image segmentation, to predict answers to simple questions about images. Our model performs 1.8 times better than the only published results on an existing image QA dataset. We also present a question generation algorithm that converts image descriptions, which are widely available, into QA form. We used this algorithm to produce an order-of-magnitude larger dataset, with more evenly distributed answers. A suite of baseline results on this new dataset are also presented.

Community Question Answering (CQA) websites are a popular tool for internet users to fulfill diverse information needs. Posted questions can be multiple sentences long and span diverse domains. They go beyond factoid questions and can be conversational, opinion-seeking and experiential questions, that might have multiple, potentially conflicting, useful answers from different users. In this paper, we describe a large-scale formative study to collect commonsense properties of questions and answers from 18 diverse communities from stackexchange.com. We collected 50,000 human judgments on 500 question-answer pairs. Commonsense properties are features that humans can extract and characterize reliably by using their commonsense knowledge and native language skills, and no special domain expertise is assumed. We report results and suggestions for designing human computation tasks for collecting commonsense semantic judgments.

The ability to perform reasoning on inconsistent data is a central problem both for AI and database research. One approach to deal with this situation is consistent query answering, where queries are answered over all possible repairs of the database. In general, the repair may be very distant from the original database. In this work we present a new approach where this distance is bounded and analyze its computational complexity. Our results show that in many (but not all) cases the complexity drops.

The standard way of answering queries over incomplete databases is to compute certain answers, defined as the intersection of query answers on all complete databases that the incomplete database represents. But is this universally accepted definition correct? We argue that this ``one-size-fits-all'' definition can often lead to counterintuitive or just plain wrong results, and propose an alternative framework for defining certain answers. We combine three previously used approaches, based on the semantics and representation systems, on ordering incomplete databases in terms of their informativeness, and on viewing databases as knowledge expressed in a logical language, to come up with a well justified and principled notion of certain answers. Using it, we show that for queries satisfying some natural conditions (like not losing information if a more informative input is given), computing certain answers is surprisingly easy, and avoids the complexity issues that have been associated with the classical definition.

The vote mechanism is widely utilized to rank answers in community-based question answering sites. In generating a vote, a user's attention is influenced by the answer position and appearance, in addition to real answer quality. Previously, these biases are ignored. As a result, the top answers obtained from this mechanism are not reliable, if the number of votes for the active question is not sufficient. In this paper, we solve this problem by analyzing two kinds of biases; position bias and appearance bias. We identify the existence of these biases and propose a joint click model for dealing with both of them. Our experiments in real data demonstrate how the ranking performance of the proposed model outperforms traditional methods with biases ignored by 15.1% in precision@1, and 11.7% in the mean reciprocal rank. A case study on a manually labeled dataset futher supports the effectiveness of the proposed model.

With Community Question Answering (CQA) evolving into a quite popular method for information seeking and providing, it also becomes a target for spammers to disseminate promotion campaigns. Although there are a number of quality estimation efforts on the CQA platform, most of these works focus on identifying and reducing low-quality answers, which are mostly generated by impatient or inexperienced answerers. However, a large number of promotion answers appear to provide high-quality information to cheat CQA users in future interactions. Therefore, most existing quality estimation works in CQA may fail to detect these specially designed answers or question-answer pairs. In contrast to these works, we focus on the promotion channels of spammers, which include (shortened) URLs, telephone numbers and social media accounts. Spammers rely on these channels to connect to users to achieve promotion goals so they are irreplaceable for spamming activities. We propose a propagation algorithm to diffuse promotion intents on an "answerer-channel" bipartite graph and detect possible spamming activities. A supervised learning framework is also proposed to identify whether a QA pair is spam based on propagated promotion intents. Experimental results based on more than 6 million entries from a popular Chinese CQA portal show that our approach outperforms a number of existing quality estimation methods for detecting promotion campaigns on both the answer level and QA pair level.

Retrieving similar questions is very important in community-based question answering. A major challenge is the lexical gap in sentence matching. In this paper, we propose a convolutional neural tensor network architecture to encode the sentences in semantic space and model their interactions with a tensor layer. Our model integrates sentence modeling and semantic matching into a single model, which can not only capture the useful information with convolutional and pooling layers, but also learn the matching metrics between the question and its answer. Besides, our model is a general architecture, with no need for the other knowledge such as lexical or syntactic analysis. The experimental results shows that our method outperforms the other methods on two matching tasks.

How should we gather information in a network, where each node's visibility is limited to its local neighborhood? This problem arises in numerous real-world applications, such as surveying and task routing in social networks, team formation in collaborative networks and experimental design with dependency constraints. Often the informativeness of a set of nodes can be quantified via a submodular utility function. Existing approaches for submodular optimization, however, require that the set of all nodes that can be selected is known ahead of time, which is often unrealistic. In contrast, we propose a novel model where we start our exploration from an initial node, and new nodes become visible and available for selection only once one of their neighbors has been chosen. We then present a general algorithm \elgreedy for this problem, and provide theoretical bounds on its performance dependent on structural properties of the underlying network. We evaluate our methodology on various simulated problem instances as well as on data collected from social question answering system deployed within a large enterprise.

How should we gather information in a network, where each node's visibility is limited to its local neighborhood? This problem arises in numerous real-world applications, such as surveying and task routing in social networks, team formation in collaborative networks and experimental design with dependency constraints. Often the informativeness of a set of nodes can be quantified via a submodular utility function. Existing approaches for submodular optimization, however, require that the set of all nodes that can be selected is known ahead of time, which is often unrealistic. In contrast, we propose a novel model where we start our exploration from an initial node, and new nodes become visible and available for selection only once one of their neighbors has been chosen. We then present a general algorithm NetExp for this problem, and provide theoretical bounds on its performance dependent on structural properties of the underlying network. We evaluate our methodology on various simulated problem instances as well as on data collected from social question answering system deployed within a large enterprise.

Identifying a patient’s important medical problems requires broad and deep medical expertise, as well as significant time to gather all the relevant facts from the patient’s medical record and assess the clinical importance of the facts in reaching the final conclusion. A patient’s medical problem list is by far the most critical information that a physician uses in treatment and care of a patient. In spite of its critical role, its curation, manual or automated, has been an unmet need in clinical practice. We developed a machine learning technique in IBM Watson to automatically generate a patient’s medical problem list. The machine learning model uses lexical and medical features extracted from a patient’s record using NLP techniques. We show that the automated method achieves 70% recall and 67% precision based on the gold standard that medical experts created on a set of de-identified patient records from a major hospital system in the US. To the best of our knowledge this is the first successful machine learning/NLP method of extracting an open-ended patient’s medical problems from an Electronic Medical Record (EMR). This paper also contributes a methodology for assessing accuracy of a medical problem list generation technique.