jurafsky&martin_3rdEd_17 (1).pdf

These questions both draws on and extends the state

Info icon This preview shows pages 400–402. Sign up to view the full content.

View Full Document Right Arrow Icon
these questions both draws on and extends the state of the art in practical question answering, as we will see. Most current question answering systems focus on factoid questions . Factoid questions are questions that can be answered with simple facts expressed in short text answers. The following factoid questions, for example, can be answered with a short string expressing a personal name, temporal expression, or location: (27.1) Who founded Virgin Airlines? (27.2) What is the average age of the onset of autism? (27.3) Where is Apple Computer based? In this chapter we describe the two major modern paradigms to question answer- ing, focusing on their application to factoid questions. The first paradigm is called IR-based question answering or sometimes text- based question answering , and relies on the enormous amounts of information available as text on the Web or in specialized collections such as PubMed. Given a user question, information retrieval techniques extract passages directly from these documents, guided by the text of the question. The method processes the question to determine the likely answer type (often a named entity like a person, location, or time), and formulates queries to send to a search engine. The search engine returns ranked documents which are broken up into suitable passages and reranked. Finally candidate answer strings are extracted from the passages and ranked. 1 The answer, of course, is Bram Stoker, and the novel was the fantastically Gothic Dracula .
Image of page 400

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
27.1 IR- BASED F ACTOID Q UESTION A NSWERING 401 In the second paradigm, knowledge-based question answering , we instead build a semantic representation of the query. The meaning of a query can be a full predicate calculus statement. So the question What states border Texas? —taken from the GeoQuery database of questions on U.S. Geography (Zelle and Mooney, 1996) — might have the representation: l x . state ( x ) ^ borders ( x , texas ) Alternatively the meaning of a question could be a single relation between a known and an unknown entity. Thus the representation of the question When was Ada Lovelace born? could be birth-year (Ada Lovelace, ?x) . Whatever meaning representation we choose, we’ll be using it to query databases of facts. These might be complex databases, perhaps of scientific facts or geospatial information, that need powerful logical or SQL queries. Or these might be databases of simple relations, triple stores like Freebase or DBpedia introduced in Chapter 20. triple stores Large practical systems like the DeepQA system in IBM’s Watson generally are hybrid systems, using both text datasets and structured knowledge bases to answer questions. DeepQA extracts a wide variety of meanings from the question (parses, relations, named entities, ontological information), and then finds large numbers of candidate answers in both knowledge bases and in textual sources like Wikipedia or newspapers. Each candidate answer is then scored using a wide variety of knowl- edge sources, such as geospatial databases, temporal reasoning, taxonomical classi- fication, and various textual sources.
Image of page 401
Image of page 402
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern