Numeric Query Answering on the Web

Numeric Query Answering on the Web

Steven O’Hara (University of Texas at San Antonio, USA) and Tom Bylander (University of Texas at San Antonio, USA)
Copyright: © 2013 |Pages: 16
DOI: 10.4018/978-1-4666-3610-1.ch011

Abstract

Query answering usually assumes that the asker is looking for a single correct answer to the question. When retrieving a textual answer this is often the case, but when searching for numeric answers, there are additional considerations. In particular, numbers often have units associated with them, and the asker may not care whether the raw answer is in feet or meters. Also, numbers usually denote a precision. In a few cases, the precision may be explicit, but normally, there is an implied precision associated with every number. Finally, an association between different reliability levels to different sources can be made. In this paper, the authors experimentally show that, in the context of conflicting answers from multiple sources, numeric query accuracy can be improved by taking advantage of units, precision, and the reliability of sources.
Chapter Preview
Top

Hypothesis

Numeric query answering is different than ordinary query answering (Katz et al., 2002; Kwok, Etzioni, & Weld, 2001; Lin & Katz, 2003; Roussinov, Fan, & Robles-Flores, 2008) in two fundamental ways: unit conversions and implied precision. Query accuracy can be improved by taking advantage of these distinct differences. It can also take advantage of source reliability, which is considered in this paper.

Numeric Values Often Have Units

When searching for numeric values, many of them will be expressed in units, such as feet or meters. When a candidate value is found, the units can be converted automatically. So, if you are searching for the radius of Mars, some answers may come back in meters, some in miles, and some in kilometers. If the search is expecting only miles, it may miss some good answers that were expressed in kilometers, for example.

Numbers Have Implicit Precision

In a few cases, numbers have explicit precision. Wikipedia, for example, often shows values in the form “2100 ± 40 grams” which means 2060 to 2140 grams. However, most resources will report values like “2100 grams” or “2110 grams”, etc. Based on common interpretation of numbers ending in zeros, we assume that the 2100 value has an implied precision of 2050 to 2150, and that the 2110 value means 2105 to 2115. These interval values can be compared against each other to see if they reinforce each other.

Naturally, this assumption can fail in both directions. One might find a value such as “1000 meters” that really means exactly 1000.000 meters. And one might find a value such as “123.45 miles” that is not really accurate to one hundredth of a mile. Part of this research is testing to see if overall accuracy improves with this assumption, even knowing that it will occasionally fail.

Complete Chapter List

Search this Book:
Reset