The Semantics of Search - Social Semantics: The Search for Meaning on the Web - page 152

x 10 5

5

4.5

4

3.5

3

2.5

2

1.5

1

0.5

0

500 1000 1500 2000 2500 3000

Popularity−ordered Queries

Fig. 6.2 The rank-ordered popularity of the queries is on the x -axis, with the y axis displaying the
number of semantic web URIs returned, with the entity queries given on bottom and the concept
queries on top

Semantic Web. The Semantic Web is not irrelevant to ordinary users as there is
data on the Semantic Web ordinary users are interested in, even if it is distributed
unevenly and does not correlate with the popularity of their queries.

6.2.2 Selecting Queries for Evaluation

In order to select a subset of informational queries for evaluation, we randomly
selected 100 queries identified as abstract concepts by WordNet and then 100
queries identified as either people or places by the named entity recognizer, for a
total of 200 queries to be used in evaluation. Constraints were placed on the URIs
resulting from semantic search, such that at least 10 Semantic Web documents (a
file containing a valid RDF graph) had to be retrieved from the URI returned by the
Semantic Web search engine. This was necessary as some queries returned 0 or less
than 10 URIs, as explained in Sect. 6.2.1 . For each query, hypertext search always
returned more than 10 URIs. So for each query, 10 Semantic Web documents were
retrieved using the FALCON-S Object Search engine (Cheng et al. 2008), leading
to a total of 1,000 Semantic Web documents about entities and 1,000 Semantic