Natural Language Questions Research Articles

Question Answering Over Knowledge Base: A Scheme for Integrating Subject and the Identified Relation to Answer Simple Questions

Answering natural language question over a knowledge base is an important and challenging task with a wide range of application in natural language processing and information retrieval. Several existing knowledge-based question answering systems exploit complex end-to-end neural network approaches that are computationally expensive and take long to execute when training the neural network. More importantly, such an end-to-end approach makes it difficult to examine the process of query processing. In this study, we decompose the question answering problem in a three-step pipeline of entity detection, entity linking, and relation prediction, and solve each component separately. We explore basic neural network and non-neural network methods for entity detection and relation prediction plus a few heuristics for entity linking. We also introduce a method to identify ambiguity in the data and show that ambiguity in the data bounds the performance of the question answering system. The experiment on the SimpleQuestions benchmark data set shows that a combination of basic LSTMs, GRUs, and non-neural network techniques achieve reasonable performance while providing an opportunity to understand the question answering problem structure.

Querying knowledge graphs in natural language

Knowledge graphs are a powerful concept for querying large amounts of data. These knowledge graphs are typically enormous and are often not easily accessible to end-users because they require specialized knowledge in query languages such as SPARQL. Moreover, end-users need a deep understanding of the structure of the underlying data models often based on the Resource Description Framework (RDF). This drawback has led to the development of Question-Answering (QA) systems that enable end-users to express their information needs in natural language. While existing systems simplify user access, there is still room for improvement in the accuracy of these systems. In this paper we propose a new QA system for translating natural language questions into SPARQL queries. The key idea is to break up the translation process into 5 smaller, more manageable sub-tasks and use ensemble machine learning methods as well as Tree-LSTM-based neural network models to automatically learn and translate a natural language question into a SPARQL query. The performance of our proposed QA system is empirically evaluated using the two renowned benchmarks-the 7th Question Answering over Linked Data Challenge (QALD-7) and the Large-Scale Complex Question Answering Dataset (LC-QuAD). Experimental results show that our QA system outperforms the state-of-art systems by 15% on the QALD-7 dataset and by 48% on the LC-QuAD dataset, respectively. In addition, we make our source code available.

Neural machine translating from natural language to SPARQL

SPARQL is a highly powerful query language for an ever-growing number of resources and knowledge graphs represented in the Resource Description Framework (RDF) data format. Using it requires a certain familiarity with the entities in the domain to be queried as well as expertise in the language’s syntax and semantics, none of which average human web users can be assumed to possess. To overcome this limitation, automatically translating natural language questions to SPARQL queries has been a vibrant field of research. However, to this date, the vast success of deep learning methods has not yet been fully propagated to this research problem. This paper contributes to filling this gap by evaluating the utilization of eight different Neural Machine Translation (NMT) models for the task of translating from natural language to the structured query language SPARQL. While highlighting the importance of high-quantity and high-quality datasets, the results show a dominance of a Convolutional Neural Network (CNN)-based architecture with a Bilingual Evaluation Understudy (BLEU) score of up to 98 and accuracy of up to 94%.

Visual Question Answering with External Knowledge

Visual Question Answering (VQA) system is an algorithm that takes an image, and a natural language question about it as input and produces an answer as the output. One of the most intriguing features of the VQA task is the uncertainty of the questions asked. To extract the knowledge required in answering them needs a spread of image manipulations. A significant part of the ongoing advancement in vision to language issues is accomplished by merging Recurrent Neural Networks and Convolutional Neural Networks. We propose a fully trainable neural network based on attribute extraction, established on the CNN and RNN design, which will apply to different vision to language issues. We do that by embedding a particular attribute representation of the scene interpretable by users. Each semantic feature corresponds to a word strip-mined from the training image description and signifies high-level information regarding the image context. We train an RCNN-based classifier for every attribute, and therefore the set of attribute likelihoods for a picture that forms a high-level representation of image content are learned. We then train an LSTM to get the captions, and answers, on the idea of these likelihoods. In light of the proposed feature-based vision to language model, the subsequent contribution is to present a means of incorporating information outside to the picture into the VQA task.

SQLSketch: Generating SQL Queries using a sketch-based approach

In the last decade, many intelligent interfaces and layers have been suggested to allow the use of relational databases and extraction of the content using only the natural language. However most of them struggle when exposed to new databases. In this article, we present SQLSketch, a sketch-based network for generating SQL queries to address the problem of automatically translate Natural Languages questions to SQL using the related databases schemas. We argue that the previous models that use full or partial sequence-to-sequence structure in the decoding phase can, in fact, have counter-effect on the generation operation and came up with more loss of the context or the meaning of the user question. In this regard, we use a full sketch-based structure that decouples the generation process into many small prediction modules. The SQLSketch is evaluated against GreatSQL, a new cross-domain, large-scale and balanced dataset for the Natural Language to SQL translation task. For a long-term aim of making better models and contributing in adding more improvements to the semantic parsing tasks, we propose the GreatSQL dataset as the first balanced cross-domain corpus that includes 45,969 pairs of natural language questions and their corresponding SQL queries in addition to simplified and well structured ground-truth annotations. We establish results for SQLSketch using GreatSQL dataset and compare the performance against two popular types of models that represent the sequential and partial-sketch based approaches. Experimental result shows that SQLSketch outperforms the baseline models by 13% in exact matching accuracy and achieve a score of 23.9% to be the new state-of-the-art model on GreatSQL.

Building natural language responses from natural language questions in the spatio-temporal context

Building natural language responses from natural language questions in the spatio-temporal context

Construction of intelligent query system for metro electromechanical equipment faults based on the knowledge graph

The intelligent inquiry system for metro electro-mechanical equipment faults based on the knowledge graph can effectively consolidate various semi-structured failure messages, and can provide users with quick, accurate and high-quality intelligent inquiry services such as equipment fault causes-researching and solutions-delivering, which could be really relevant to this research field and application areas. The recorded date which related to metro electromechanical equipment failures were in this research collected, consolidated and converted, so that these failures could be stored in our databases. In this context, various functions of the intelligent inquiry system have been implemented, including: natural language question analysis, language Cypher-based question and answer design, Naive Bayesian classification based on characteristic core words, and user interaction interface realization. The experimental results show that the system can effectively solve the problems related to fault handling in metro mechanical and electrical equipment, thus improving the efficiency of equipment fault maintenance.

Improved Entity Linking for Simple Question Answering Over Knowledge Graph

Question Answering systems over Knowledge Graphs (KG) answer natural language questions using facts contained in a knowledge graph, and Simple Question Answering over Knowledge Graphs (KG-SimpleQA) means that the question can be answered by a single fact. Entity linking, which is a core component of KG-SimpleQA, detects the entities mentioned in questions, and links them to the actual entity in KG. However, traditional methods ignore some information of entities, especially entity types, which leads to the emergence of entity ambiguity problem. Besides, entity linking suffers from out-of-vocabulary (OOV) problem due to the limitation of pre-trained word embeddings. To address these problems, we encode questions in a novel way and encode the features contained in the entities in a multilevel way. To evaluate the enhancement of the whole KG-SimpleQA brought by our improved entity linking, we utilize a relatively simple approach for relation prediction. Besides, to reduce the impact of losing the feature during the encoding procedure, we utilize a ranking algorithm to re-rank (entity, relation) pairs. According to the experimental results, our method for entity linking achieves an accuracy of 81.8% that beats the state-of-the-art methods, and our improved entity linking brings a boost of 5.6% for the whole KG-SimpleQA.

Comparison of Services for Creating Question-Answering Systems

Question-answering systems are widely used nowadays. They allow automating interaction between people. Common solutions include client support, recommendations, and order processing.A question-answering system is a system that answers natural language questions using a structured database or natural language document collection. It can be considered as the development of information retrieval.Three are different services for creating question-answering systems. In this work selected Dialog- flow, IBM Watson Assistant, Microsoft QnA Maker, and LUIS. Each service has its advantages and disadvantages.The main concepts of Dialogflow are agents, intents, entities, context. An agent is a virtual agent with a natural language understanding that handles conversations with users. An intent represents user intention for conversation turn. Each agent has its intents. Entities are parameters of intent that Dialogflow extracts from user expressions. A context in Dialogflow similar to natural language context and helps to control the flow of conversation.The main concepts of IBM Watson Assistant are intents and entities. Intents represent users’ goals. User inputs analyzed to select the most appropriate intent and choose dialog flow. An entity represents information in intent that is relevant to the user’s goal.Microsoft QnA Maker uses a concept of a knowledge base that is filled with information. Requests from the user are analyzed to extract the most appropriate information from the knowledge base. LUIS uses concepts of intent and entities. LUIS uses schemas to extract intent or entity from an input.Important for the choice are the tasks to be solved by the created question-answer system and the volume of system use. Each service has restrictions on the languages that are supported. Dialogflow supports 19 languages, IBM Watson Assistant only 10, Microsoft QnA Maker 52 languages, but LUIS only 17. All services support English, but only Dialogflow and Microsoft QnA Maker support Ukrainian and Russian. .Each service has standard integrations with messengers. All services support Slack and Facebook Messenger. The rest have different support from different services. All services provide options to create a custom integration for non-standard systems if the service does not provide them, through the appropriate APIs.Prices depend on the number of requests per month with a possible limit on the number per minute. Each service offers a free plan under certain conditions, but Microsoft services will require payment for Azure capacity.Manuscript received 09.06.2020

Learning to generate structured queries from natural language with indirect supervision

Generating structured query language (SQL) from natural language is an emerging research topic. This paper presents a new learning paradigm from indirect supervision of the answers to natural language questions, instead of SQL queries. This paradigm facilitates the acquisition of training data due to the abundant resources of question-answer pairs for various domains in the Internet, and expels the difficult SQL annotation job. An end-to-end neural model integrating with reinforcement learning is proposed to learn SQL generation policy within the answer-driven learning paradigm. The model is evaluated on datasets of different domains, including movie and academic publication. Experimental results show that our model outperforms the baseline models.

Break It Down: A Question Understanding Benchmark

Understanding natural language questions entails the ability to break down a question into the requisite steps for computing its answer. In this work, we introduce a Question Decomposition Meaning Representation (QDMR) for questions. QDMR constitutes the ordered list of steps, expressed through natural language, that are necessary for answering a question. We develop a crowdsourcing pipeline, showing that quality QDMRs can be annotated at scale, and release the Break dataset, containing over 83K pairs of questions and their QDMRs. We demonstrate the utility of QDMR by showing that (a) it can be used to improve open-domain question answering on the HotpotQA dataset, (b) it can be deterministically converted to a pseudo-SQL formal language, which can alleviate annotation in semantic parsing applications. Last, we use Break to train a sequence-to-sequence model with copying that parses questions into QDMR structures, and show that it substantially outperforms several natural baselines.

Improving Access to Scientific Literature with Knowledge Graphs

AbstractThe transfer of knowledge has not changed fundamentally for many hundreds of years: It is usually document-based-formerly printed on paper as a classic essay and nowadays as PDF. With around 2.5 million new research contributions every year, researchers drown in a flood of pseudo-digitized PDF publications. As a result research is seriously weakened. In this article, we argue for representing scholarly contributions in a structured and semantic way as a knowledge graph. The advantage is that information represented in a knowledge graph is readable by machines and humans. As an example, we give an overview on the Open Research Knowledge Graph (ORKG), a service implementing this approach. For creating the knowledge graph representation, we rely on a mixture of manual (crowd/expert sourcing) and (semi-)automated techniques. Only with such a combination of human and machine intelligence, we can achieve the required quality of the representation to allow for novel exploration and assistance services for researchers. As a result, a scholarly knowledge graph such as the ORKG can be used to give a condensed overview on the state-of-the-art addressing a particular research quest, for example as a tabular comparison of contributions according to various characteristics of the approaches. Further possible intuitive access interfaces to such scholarly knowledge graphs include domain-specific (chart) visualizations or answering of natural language questions.

WabiQA: A Wikipedia-Based Thai Question-Answering System

With vast information that has been digitized and made available online, manually finding the answer to a question can be tedious. While search engines have emerged to facilitate information needs, users would have to manually read through the retrieved articles to locate the answer to a specific question. Therefore, the ability to automatically understand users’ natural language questions and find the correct answers could prove crucial in information retrieval. Indeed, such automatic question-answering solutions have been extensively studied by the natural language processing (NLP) research communities. However, most of the development targets questions and information sources composed in high-resource languages such as English and Chinese. In this paper, we propose WabiQA, a novel system for automatically answering questions in the Thai language using the Thai Wikipedia articles as the knowledge source. Specifically, the proposed method first retrieves the Wikipedia article that is most likely to contain the answer. Then, a bi-directional LSTM model is used to read the article and locate candidate answers, which are ranked by confidence levels and returned to the user. WabiQA won the first prize award from Thailand’s National Software Contest 2019 under category “Question-Answering Program from Thai Wikipedia,” with 83.5%, 34.80%, and 45.96%, and outperforming the next best competitors’ systems by 19.99, 24.26, and 33.10 percentage points in terms of Accuracy@1, EM, and F1 respectively. Furthermore, we also develop a prototype mobile application that aims to facilitate Thai users with visual impairment using voice-to-speech technology and an intelligent question-answer categorization. The findings of this research not only expand the horizon of the possibility to develop intelligent NLP applications for the Thai language using only available existing Thai NLP tools, resources, and deep learning technologies, but also shed light on the possibility to apply such techniques to develop many intelligent NLP tasks for the Thai and other low-resource languages such as reading assessment, writing assistance, and entity linking.

Building natural language responses from natural language questions in the spatio-temporal context

With the evolving research in geographic information system (GIS) owing to its ability to support decision makers in different fields, there is a strong need to enabling all users; specialists and non-specialists to profit from this technology. Although, the key impediment to non-specialists is the language to interact with the GIS and especially its embedded geographic database (GDB) which require SQL skills. In this paper we explore a new approach which alleviates nomad GIS users from any formatting effort by only using the natural language as a GDB communication mean. The process is generally two-fold: 1) formatting the natural language user query to be processed by the GDB engine; 2) translating the GDB retrieved answer to a text easily interpreted by all GIS users. The resulting implemented system was integrated to the OpenJump GIS and has been evaluated to give satisfactory results.

Trainable Framework for Information Extraction, Structuring and Summarization of Unstructured Data, Using Modified NER

The World Wide Web is an ever expanding source of data in today’s world. Millions of tera-bytes of data and information is getting added every second. In this information age as the data is getting generated at an exponential rate, the fact to be noted is that most of the information is already available is in the form of natural language text. The task of information extraction from mammoth data leads us to think on the quality and the form of available data. Secondly, the ever increasing data poses a challenging task of extracting useful information from the available data. The third task is to extract information as efficiently as possible. For retrieving the information there is a need to develop ingenious way to answer any kind of query put up by a user from the available unstructured data. This paper proposes a novel trainable and integrated Natural Language Information Interpretation and Representation System (NLIIRS) that accepts any available un-annotated corpus of data in the form of natural language, and performs the following tasks: finds out the useful data, extracts relevant information in usable form (structured form/tables), summarizes the data and structures the data in relational form. At the end the Question and Answering (Q&A) module shows the cognitive abilities of NLIIR system by answering the questions in natural language relevant to the text. This multispecialty system beyond just Q&A. This is a trainable system capable of handling any unstructured data to be transformed into structured and well organized information. It allows the user to ask questions in natural language. It adopts the advantages of a modified named entity recognition so as to bypass the time consuming process of parts of speech tagging while pre-processing the available corpus (data) for information extraction.

Visual Experience-Based Question Answering with Complex Multimodal Environments

This paper proposes a novel visual experience-based question answering problem (VEQA) and the corresponding dataset for embodied intelligence research that requires an agent to do actions, understand 3D scenes from successive partial input images, and answer natural language questions about its visual experiences in real time. Unlike the conventional visual question answering (VQA), the VEQA problem assumes both partial observability and dynamics of a complex multimodal environment. To address this VEQA problem, we propose a hybrid visual question answering system, VQAS, integrating a deep neural network-based scene graph generation model and a rule-based knowledge reasoning system. The proposed system can generate more accurate scene graphs for dynamic environments with some uncertainty. Moreover, it can answer complex questions through knowledge reasoning with rich background knowledge. Results of experiments using a photo-realistic 3D simulated environment, AI2-THOR, and the VEQA benchmark dataset prove the high performance of the proposed system.

Visual Question Answering for Cultural Heritage

Technology and the fruition of cultural heritage are becoming increasingly more entwined, especially with the advent of smart audio guides, virtual and augmented reality, and interactive installations. Machine learning and computer vision are important components of this ongoing integration, enabling new interaction modalities between user and museum. Nonetheless, the most frequent way of interacting with paintings and statues still remains taking pictures. Yet images alone can only convey the aesthetics of the artwork, lacking is information which is often required to fully understand and appreciate it. Usually this additional knowledge comes both from the artwork itself (and therefore the image depicting it) and from an external source of knowledge, such as an information sheet. While the former can be inferred by computer vision algorithms, the latter needs more structured data to pair visual content with relevant information. Regardless of its source, this information still must be be effectively transmitted to the user. A popular emerging trend in computer vision is Visual Question Answering (VQA), in which users can interact with a neural network by posing questions in natural language and receiving answers about the visual content. We believe that this will be the evolution of smart audio guides for museum visits and simple image browsing on personal smartphones. This will turn the classic audio guide into a smart personal instructor with which the visitor can interact by asking for explanations focused on specific interests. The advantages are twofold: on the one hand the cognitive burden of the visitor will decrease, limiting the flow of information to what the user actually wants to hear; and on the other hand it proposes the most natural way of interacting with a guide, favoring engagement.

Better Modeling the Hierarchical Structure of Language for Sentence Semantic Matching

Sentence semantic matching requires understanding the semantic relationship between two sentences, which is widely used in various natural language tasks such as paraphrase identification, natural language inference and question answering. Usually, the underlying structure of sentence is not strictly sequential, this structure is hierarchical. This paper studies how to better model the hierarchical structure of sentence for semantic matching. Recent research suggests that BERT implicitly captures classical, tree-like structures. Based on this, this paper proposes to further enhance the strength of modeling hierarchical structure of language with an advanced variant of LSTM – Ordered Neurons LSTM(ON-LSTM), which introduces a syntax-oriented inductive bias to perform the composition of classical tree structure. Experimental results demonstrate that the proposed approach, which enhances hierarchical structure of language, significantly improves the performance of sentence semantic matching.

Less is more: Data-efficient complex question answering over knowledge bases

Question answering is an effective method for obtaining information from knowledge bases (KB). In this paper, we propose the Neural-Symbolic Complex Question Answering (NS-CQA) model, a data-efficient reinforcement learning framework for complex question answering by using only a modest number of training samples. Our framework consists of a neural generator and a symbolic executor that, respectively, transforms a natural-language question into a sequence of primitive actions, and executes them over the knowledge base to compute the answer. We carefully formulate a set of primitive symbolic actions that allows us to not only simplify our neural network design but also accelerate model convergence. To reduce search space, we employ the copy and masking mechanisms in our encoder–decoder architecture to drastically reduce the decoder output vocabulary and improve model generalizability. We equip our model with a memory buffer that stores high-reward promising programs. Besides, we propose an adaptive reward function. By comparing the generated trial with the trials stored in the memory buffer, we derive the curriculum-guided reward bonus, i.e., the proximity and the novelty. To mitigate the sparse reward problem, we combine the adaptive reward and the reward bonus, reshaping the sparse reward into dense feedback. Also, we encourage the model to generate new trials to avoid imitating the spurious trials while making the model remember the past high-reward trials to improve data efficiency. Our NS-CQA model is evaluated on two datasets: CQA, a recent large-scale complex question answering dataset, and WebQuestionsSP, a multi-hop question answering dataset. On both datasets, our model outperforms the state-of-the-art models. Notably, on CQA, NS-CQA performs well on questions with higher complexity, while only using approximately 1% of the total training samples.

Visual Question Answering Based on Question Attention Model

Visual Question Answer (VQA), the natural language question of Visual images, has become popular in the field of artificial intelligence. At present, most of the VQA models extract the whole image features, which consume a large amount of computation and have a complex structure. In this paper, we propose a VQA method based on question attention model. Firstly, the Convolutional Neural Networks (CNN) is used to extract image features from the input images, and the question text is processed by the Long Short-Term Memory (LSTM). Then, we design a question attention module to let the learning algorithm focus on the most relevant features of the input text. According to question features, our method utilities the attention module to add the corresponding weights to the image features and extract the meaningful information for the generation of answer sequence words. Our method performed significantly better than the LSTMQ+I model on the MS COCO visual question answer (VQA) dataset with an accuracy improvement of 2%.