Ways of Forming Query Narratives¶

In this section we will consider how to improve the precision of a query narrative by first examining aspects of these narratives that can be misleading. Next we will see examples of narratives that show the importance of being sensitive to the distinction between singular and plural.

!!!!!!!!!!!!!!! Fix after completing

Misleading Narratives¶

When you are working to express queries in narrative form, there are several kinds of misleading narratives that you should look out for. They are: missing domain knowledge, assumed context, missing qualifiers, almost identifiers, and vague aggregates.

Missing domain knowledge. Communication between a user and an analyst can fail. Sometimes misunderstanding can stem from unfamiliar vocabulary. For example, as an analyst who is not completely familiar with the creature database and what the proficiency values represent, you might be given the following task:

Find each poor floater.

There are a couple of things that your client left out of this request. First, we need to pin down what a floater is. Is it a creature who has achieved the float skill? Or if we are not familiar with whether float is a team skill or whether a creature can contribute to the float achievement, we would need to pin that down. Second, the client has failed to state what constitutes poor floating. We would need to know what level of proficiency would mean ‘poor’. With this kind of data, ‘poor’ is likely a different value for each skill. This type of domain knowledge is not given in this short request.

Other surprises show up when worlds collide. For example, computer scientists generally use “process” as a verb (process data), but in biology “process” sometimes is used as a noun, a place, such as the xyphoid process (at the bottom of the sternum). Also, sometimes misunderstanding can stem from users inventing new uses for words. For example, primatologists record data as they follow chimps in their native habitat. Although follow is a verb, they sometimes use it as a noun, as in “the January 2, 1977 follow”.