Semantic matching in machine reading comprehension: An empirical study

Machine reading comprehension (MRC) is a challenging task in the field of artificial intelligence. Most existing MRC works contain a semantic matching module, either explicitly or intrinsically, to determine whether a piece of context answers a question. However, there is scant work which systematically evaluates different paradigms using semantic matching in MRC. In this paper, we conduct a systematic empirical study on semantic matching. We formulate a two-stage framework which consists of a semantic matching model and a reading model, based on pre-trained language models. We compare and analyze the effectiveness and efficiency of using semantic matching modules with different setups on four types of MRC datasets. We verify that using semantic matching before a reading model improves both the effectiveness and efficiency of MRC. Compared with answering questions by extracting information from concise context, we observe that semantic matching yields more improvements for answering questions with noisy and adversarial context. Matching coarse-grained context to questions, e.g., paragraphs, is more effective than matching fine-grained context, e.g., sentences and spans. We also find that semantic matching is helpful for answering who/where/when/what/how/which questions, whereas it decreases the MRC performance on why questions. This may imply that semantic matching helps to answer a question whose necessary information can be retrieved from a single sentence. The above observations demonstrate the advantages and disadvantages of using semantic matching in different scenarios.     

Question categorization and classification using grammar based approach

Question-answering has become one of the most popular information retrieval applications. Despite that most question-answering systems try to improve the user experience and the technology used in finding relevant results, many difficulties are still faced because of the continuous increase in the amount of web content. Questions Classification (QC) plays an important role in question-answering systems, with one of the major tasks in the enhancement of the classification process being the identification of questions types. A broad range of QC approaches has been proposed with the aim of helping to find a solution for the classification problems; most of these are approaches based on bag-of-words or dictionaries. In this research, we present an analysis of the different type of questions based on their grammatical structure. We identify different patterns and use machine learning algorithms to classify them. A framework is proposed for question classification using a grammar-based approach (GQCC) which exploits the structure of the questions. Our findings indicate that using syntactic categories related to different domain-specific types of Common Nouns, Numeral Numbers and Proper Nouns enable the machine learning algorithms to better differentiate between different question types. The paper presents a wide range of experiments the results show that the GQCC using J48 classifier has outperformed other classification methods with 90.1% accuracy.     

Addressing ontology-based question answering with collections of user queries

This paper presents QACID an ontology-based Question Answering system applied to the CInema Domain. This system allows users to retrieve information from formal ontologies by using as input queries formulated in natural language. The original characteristic of QACID is the strategy used to fill the gap between users’ expressiveness and formal knowledge representation. This approach is based on collections of user queries and offers a simple adaptability to deal with multilingual capabilities, inter-domain portability and changes in user information requirements. All these capabilities permit developing Question Answering applications for actual users. This system has been developed and tested on the Spanish language and using an ontology modelling the cinema domain. The performance level achieved enables the use of the system in real environments.     

Do systems pass university entrance exams?

Reading Comprehension tests are commonly used to assess the degree to which people comprehend what they read. This is why we work with the hypothesis that it is reasonable to use these tests to assess the degree to which a machine “comprehends” what it is reading. In this work, we evaluate Question Answering systems using Reading Comprehension tests from exams to enter University. This article analyses the datasets generated, the kind of inferences required, the methodology followed in three evaluation campaigns, the approaches presented by participants and current results. Besides, we study the evolution of systems and the main lessons learned in this evaluation process. We also show how current technologies are unable to pass university-entrance exams. This is because these tests require a deep understanding of texts, as well as detecting the similar meaning of phrases with different words. Future directions focused on these ideas seem more promising than including a massive amount of data for training systems, what has allowed systems to obtain outstanding results in Reading Comprehension tests with more straightforward questions. We think this study helps to increase the knowledge about how to develop better Question Answering systems.     

A structural support vector method for extracting contexts and answers of questions from online forums

This article addresses the issue of extracting contexts and answers of questions from posts of online discussion forums. In previous work, general-purpose graphical models have been employed without any customization to this specific extraction problem. Instead, in this article, we propose a unified approach to context and answer extraction by customizing the structural support vector machine method. The customization enables our proposal to explore various relations among sentences of posts and complex structures of threads. We design new inference algorithms to find or approximate the most violated constraint by utilizing the specific structure of forum threads, which enables us to efficiently find the global optimum of the customized optimizing problem. We also optimize practical performance measures by varying loss functions. Experimental results show that our methods are both promising and flexible.     

Using lexical chains for keyword extraction

Keywords can be considered as condensed versions of documents and short forms of their summaries. In this paper, the problem of automatic extraction of keywords from documents is treated as a supervised learning task. A lexical chain holds a set of semantically related words of a text and it can be said that a lexical chain represents the semantic content of a portion of the text. Although lexical chains have been extensively used in text summarization, their usage for keyword extraction problem has not been fully investigated. In this paper, a keyword extraction technique that uses lexical chains is described, and encouraging results are obtained.     

A knowledge-rich approach to identifying semantic relations between nominals

This paper describes a state-of-the-art supervised, knowledge-intensive approach to the automatic identification of semantic relations between nominals in English sentences. The system employs a combination of rich and varied sets of new and previously used lexical, syntactic, and semantic features extracted from various knowledge sources such as WordNet and additional annotated corpora. The system ranked first at the third most popular SemEval 2007 Task – Classification of Semantic Relations between Nominals and achieved an F-measure of 72.4% and an accuracy of 76.3%. We also show that some semantic relations are better suited for WordNet-based models than other relations. Additionally, we make a distinction between out-of-context (regular) examples and those that require sentence context for relation identification and show that contextual data are important for the performance of a noun–noun semantic parser. Finally, learning curves show that the task difficulty varies across relations and that our learned WordNet-based representation is highly accurate so the performance results suggest the upper bound on what this representation can do.     

Is my stance the same as your stance? A cross validation study of stance detection datasets

Stance detection identifies a person’s evaluation of a subject, and is a crucial component for many downstream applications. In application, stance detection requires training a machine learning model on an annotated dataset and applying the model on another to predict stances of text snippets. This cross-dataset model generalization poses three central questions, which we investigate using stance classification models on 7 publicly available English Twitter datasets ranging from 297 to 48,284 instances. (1) Are stance classification models generalizable across datasets? We construct a single dataset model to train/test dataset-against-dataset, finding models do not generalize well (avg F1=0.33). (2) Can we improve the generalizability by aggregating datasets? We find a multi dataset model built on the aggregation of datasets has an improved performance (avg F1=0.69). (3) Given a model built on multiple datasets, how much additional data is required to fine-tune it? We find it challenging to ascertain a minimum number of data points due to the lack of pattern in performance. Investigating possible reasons for the choppy model performance we find that texts are not easily differentiable by stances, nor are annotations consistent within and across datasets. Our observations emphasize the need for an aggregated dataset as well as consistent labels for the generalizability of models.     

Interactive question answering: Is Cross-Language harder than monolingual searching?

Is Cross-Language answer finding harder than Monolingual answer finding for users? In this paper we provide initial quantitative and qualitative evidence to answer this question.     

Combining evidence with a probabilistic framework for answer ranking and answer merging in question answering

Question answering (QA) aims at finding exact answers to a user’s question from a large collection of documents. Most QA systems combine information retrieval with extraction techniques to identify a set of likely candidates and then utilize some ranking strategy to generate the final answers. This ranking process can be challenging, as it entails identifying the relevant answers amongst many irrelevant ones. This is more challenging in multi-strategy QA, in which multiple answering agents are used to extract answer candidates. As answer candidates come from different agents with different score distributions, how to merge answer candidates plays an important role in answer ranking. In this paper, we propose a unified probabilistic framework which combines multiple evidence to address challenges in answer ranking and answer merging. The hypotheses of the paper are that: (1) the framework effectively combines multiple evidence for identifying answer relevance and their correlation in answer ranking, (2) the framework supports answer merging on answer candidates returned by multiple extraction techniques, (3) the framework can support list questions as well as factoid questions, (4) the framework can be easily applied to a different QA system, and (5) the framework significantly improves performance of a QA system. An extensive set of experiments was done to support our hypotheses and demonstrate the effectiveness of the framework. All of the work substantially extends the preliminary research in Ko et al. (2007a). A probabilistic framework for answer selection in question answering. In: Proceedings of NAACL/HLT.     

User interest modeling and its application for question recommendation in user-interactive question answering systems

In this paper, we propose a generative model, the Topic-based User Interest (TUI) model, to capture the user interest in the User-Interactive Question Answering (UIQA) systems. Specifically, our method aims to model the user interest in the UIQA systems with latent topic method, and extract interests for users by mining the questions they asked, the categories they participated in and relevant answer providers. We apply the TUI model to the application of question recommendation, which automatically recommends to certain user appropriate questions he might be interested in. Data collection from Yahoo! Answers is used to evaluate the performance of the proposed model in question recommendation, and the experimental results show the effectiveness of our proposed model.