Scored matching provides a way of finding relevant records and listing them in order so
that you can identify how well those records match a query. It is a useful analysis method where
several fields and values might contribute to what makes the records interesting to you. The result
of a scored match is always a list of entities of one type, which is listed in order of
score.
About this task
Scored matching relies on you assigning a score, or weighting, to fields that match
specified conditions. Those records with the highest combined score are listed first. You specify
how the score is to be applied to individual fields. For example, you might be interested in dark
haired people, so you score the Hair Color field: you weight the Black field value with 10, the Dark
Brown value with 9, down to Fair with 1.If you select a query as the source, the scored matching
considers only the results of the query that are selected in the Entity Type to match list.
Selecting a query makes it possible to score fields and values that relate to any of the other
entities and links in the query.
For a query to be used as a source of data:
- It must be saved.
- The results must include the entity type that you want to see in the results of the scored
match. You can include other entity and link types.
- If it has two outputs of the same entity type, the names used for the entities in the query
structure should be meaningful. For example, you are unlikely to understand the significance of
names such as Person 1 and Person 2 without examining the query.
- It should be as general as possible so that it does not exclude records that are unlikely
matches but where you might want to define low scores.
Procedure
To use scored matching:
-
From the Entity Type to match list, select the type of entity you want to find (the 'match
type').
-
In the Match on records from the following source area, select the scope of the search.
If you select a query, you may see Match to this item if there is a choice
of output entities for the ‘match type’. Select the entity you want to match.
Note: You cannot select a query that has a count condition on a link, such as count = 0, meaning, for
example, 'give me all the people who have no address links'.
-
Set the minimum score you require in the Minimum score box.
-
Specify conditions and the scores to apply when the conditions are matched. In each line,
specify the Field, Operator, Value
1 and Score (for some operators Value 2 is
also required). The * character indicates the selected line.
You can:
- Click an empty row to start a new scored condition or click Insert. However, the order of the
lines will have no effect on the results.
- Use the same operators as in querying. Note: in list/not in list are not available for scored
matching.
- Use wildcards in field values.
- Click Repeat to copy the selected line.
- Click Delete to remove the selected line.
If you specify a query for the source, you can also specify conditions and scores for links and
linked entities in the query. There are pages for each link and link end entity.
-
Click Results to list the records with their relevant scores on the Results page.
