Software Problems That Motivate Web Searches - Finding Source Code on the Web for Remix and Reuse - page 258

Query
Reformulation

0 1-7 Bookmark No info

Remembering/
Fact Finding

13 5 4 0

Clarifying 15 14 3 5

Opportunistic
Searches

Learning 6 6 0 0

SUBTOTAL 34 (48%) 25 (35%) 7 (10%) 5 (7%)

Looking for Open
Source Projects or
Software Tools

14 2 0 0

Non-Opportunistic
Searches

SUBTOTAL 14 (88%) 2 (13%) 0 (0%) 0 (0%)

TOTAL 48 (55%) 27 (31%) 7 (8%) 5 (6%)

Table 13.2: Query reformulation by type of software problems

One instance of a search that involved query refinement was when Manfred was
looking for a solution to solve an exception he was having when he was using Htm-
lUnit. He entered the following sequence of queries:
Query 1: htmlunit “The data necessary to complete this operation is not yet avail-
able”
Query 2: htmlunit doScroll
Query 3: htmlunit “The data necessary to complete this operation is not yet avail-
able” doscroll
In the first query, Manfred entered the name of the library and the error message
he received. After examining some results, he learned that this problem could be
related to the “doScroll” method, so he replaced the error message with the name
of the method in the second query. Then, he examined the results and did not find
the answer he was looking for, so he tried to include both the error message and the
name of the method. This example shows that developers collect partial knowledge
from the search result evaluation they perform during a Web search session.

13.4.1.2 Searches on the Web Did Not Follow a Well-Planned Process

Web searches happened in an ad hoc manner and they happen often. Typically, de-
velopers did not start their day by planning the Web searches they are going to