Database Reference
In-Depth Information
In Figure 13, we propose a classification of the techniques presented and discussed in this chapter.
This classification is based on the situation or problem, from those mentioned above, that they are sup-
posed to address.
The first category (
Group A
in Figure 13) contains query relaxation (Section 2.1.1) and similarity-
based search (Section 2.1.2) techniques. These techniques address situations in which a user approach-
es the database with a query that exactly captures her/his retrieval goal but she/he is delivered an
empty result set. These techniques allow the database system to retrieve results that closely (though not
completely) match the user's retrieval goal.
The second (
Group B
in Figure 13) contains preference-based (Section 2.3.1), fuzzy-based (Section
2.3.2) and keyword-based search (Section 2.3.3) techniques. These techniques provide human-oriented
interfaces which allow users to formulate their retrieval goals in a more natural or intuitive manner.
Note, however, that these techniques, while useful, do not help users to clarify or refine their retrieval
goals; therefore they are not designed for the problem of ill-defined retrieval goal.
Finally, the third category (
Group C
in Figure 13) contains automated ranking and clustering-based
techniques. These techniques address situations in which a database user has an ill-defined retrieval goal.
Automated ranking-based techniques (Section 2.2.1) first seek to clarify or approximate the retrieval
goal. For this purpose, they use either past behavior of the user (derived from available workloads) or
relevance feedback from the user. Then, they compute a score, by means of a similarity measure, of
each answer that represents the extent to which it is relevant to the approximated user's retrieval goal.
Finally, the user is provided with a ranked list, in descending order of relevance, of either all query
results or only a top-k subset. The effectiveness of these approaches highly depends on their ability to
accurately capture the user's retrieval goal, which is a tedious and time consuming task. Note that such
approaches also bring the disadvantage of match homogeneity, i.e., the user is often required to go through
Figure 13. A classification of advanced database query processing techniques
