Disambiguating For The Web: A Test Of Two Methods

The Semantic Web vision expects authors to represent knowledge unambiguously, but their ability and willingness to do so are contested. To evaluate experimentally two disambiguation methods that authors might use, we showed sentences from the Web containing syntactically ambiguous quantification to 386 subjects and asked them to choose between pairs of paraphrasal and/or truth-conditional restatements. The paraphrasal method was mostly superior. Subjects generally found both methods satisfying and were able to achieve substantial consistency and agreement.