Explore chapters and articles related to this topic
Applications of NLP and IR
Published in Anuradha D. Thakare, Shilpa Laddha, Ambika Pawar, Hybrid Intelligent Systems for Information Retrieval, 2023
Anuradha D. Thakare, Shilpa Laddha, Ambika Pawar
NLP and AI technology advancement has proven a lot of enhancements in its application domain to improve user experience and thus in this application of providing perfect answers for user’s natural language query/question. NLP has a lot of applications in the IR information retrieval process; it works at various layers of linguistic analysis and transforms data to get processed by machine and device intelligent solutions. Intelligent question answering is an important application of NLP and IR. A lot of QA systems are developed for finding short and precise results for users’ natural language questions. NLP techniques help machines process information like human beings and help to reduce human efforts of finding correct answers from a set of answers. NLP helps to interpret user queries and documents with accurate indexing and also minimizes the search space. TF-IDF does proper normalization to find top-ranked relevant documents for user questions; cosine similarity measure is used in this process. Beast match (BM25 Algorithm) further reduces response time and helps in retrieving relevant answers from a set of passages in the document.
Building Applications
Published in James Luke, David Porter, Padmanabhan Santhanam, Beyond Algorithms, 2022
James Luke, David Porter, Padmanabhan Santhanam
The best examples are popular Chatbots such as Apple/Siri, Amazon/Alexa, Google/Home or Microsoft/Cortana used in the simple question answering mode. Figure 2.5 represents the high-level view of a speech-based Chatbot. When the user speaks “Who is the president of the United States?”, it gets transcribed by the “Speech-to-Text” service into the corresponding text. The text gets passed as input to the next AI task “Question-Answering”, which is another service in the popular domain from the companies, which brings the answer as “Joe Biden”. The answer gets passed to the “Text-to-Speech” service, which renders the answer to the user. For simplicity, we have represented ‘Question-Answering’ as one AI component in Figure 2.5, but in reality, it will consist of many subcomponents [21]. There is a user input analysis component that extracts user intent (“Name”) and entity identification (“President of the United States”). It may also perform additional tasks such as user sentiment analysis. There is a dialogue management component that manages ambiguities, errors & information retrieval and a response generation component which prepares the textual content to be delivered to the user.
A Brief History of Artificial Intelligence
Published in Ron Fulbright, Democratization of Expertise, 2020
From 2006–2011, IBM developed Watson, a question answering computing system initially developed to answer questions on the quiz show Jeopardy! To accomplish this goal, Watson was designed to apply advanced natural language processing, information retrieval, knowledge representation, automated reasoning, and machine learning technologies (Deshpande et al., 2017). In 2011, Watson participated in the Jeopardy! Challenge and defeated legendary champions Brad Rutter and Ken Jennings (Markoff, 2011). Since 2011, IBM has developed many applications on the Watson platform across multiple domains including healthcare, teaching assistants (Leopold, 2017), weather forecasting (Jancer, 2016), tax preparation (Moscaritolo, 2017), and a chatbot providing conversation for children’s toys (Takahashi, 2015).
User Evaluation of Conversational Agents for Aerospace Domain
Published in International Journal of Human–Computer Interaction, 2023
Ying-Hsang Liu, Alexandre Arnold, Gérard Dupont, Catherine Kobus, François Lancelot, Géraud Granger, Yves Rouillard, Alexandre Duchevet, Jean-Paul Imbert, Nadine Matton
A recent survey of dialogue systems has highlighted various kinds of conversational systems associated with conversational agents, including task-oriented dialogue systems, conversational agents, and interactive question-answering systems (Deriu et al., 2021). Issues related to the voice-based user interface, such as recognition errors, UX, and voice queries, have become more prominent in the fields of human–computer interaction (HCI), UX, and information retrieval (IR). While systematic reviews of empirical user studies of conversational agents have revealed the importance of agent performance quality, such as knowledge level and task completion, for building trust in these artificial intelligence devices (Rheu et al., 2021), the efficacy and health outcomes of using these systems in the healthcare domain have rarely been evaluated (Laranjo et al., 2018). From a user or human-centered AI perspective, the usefulness of these systems in supporting user tasks has not been well-established (C. Liu et al., 2021; Shneiderman, 2022).
A survey on non-factoid question answering systems
Published in International Journal of Computers and Applications, 2022
Manvi Breja, Sanjay Kumar Jain
Accurately answering non-factoid questions is still a big challenge in the question answering domain. The paper identifies various issues and challenges faced in each module of QAS. Various techniques employed for each module of QAS with their performance metrics are categorized. Research gaps and research questions have also been highlighted which will help researchers to seek the area needed to work on to explore more optimum solutions with negligible limitations in each module of non-factoid QAS and further enhance its performance. The paper provides a clear idea of the research in non-factoid QAS which will aid researchers in investigating new techniques to improve its performance. The survey carried out helps in understanding improvements that can be addressed in future: question classification phase requires an analysis of integrated linguistic features in questions that help in labeling questions and their answer type, answer candidate extraction by identifying semantic relations and employing common sense knowledge, and answer re-ranker by finding linguistic similarities between question and answer candidates.
Deep learning based question answering system in Bengali
Published in Journal of Information and Telecommunication, 2021
Tasmiah Tahsin Mayeesha, Abdullah Md Sarwar, Rashedur M. Rahman
Question Answering (QA) involves building systems that are able to automatically respond to questions posed by humans in a natural language. Text based Question Answering tasks can be formulated as information retrieval problems where we want to find the documents that answer a certain question, extract the potential answers from the documents and rank them, or as reading comprehension problems where the task is to find the answer (also called span) from a context passage of text. While Question Answering tasks can encompass visual (context being image), open domain, multimodal (context can be image/video/audio/text) or focus on common sense reasoning tasks, in this work we focus on text-based reading comprehension. Reading comprehension models have great practical value especially for industry purposes as a properly trained reading comprehension model can function as a chatbot for answering frequently asked questions. The driver behind progress of question answering research has been the curation of multiple high-quality large reading comprehension datasets and release of models performing well on these datasets. However, for low resource languages like Bengali similar efforts to annotate large high-quality reading comprehension datasets can be costly and challenging due to lack of skilled annotators, awareness and Bengali specific NLP tools.