27.03.2014 Views

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SEKE 2012 Proceedings - Knowledge Systems Institute

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

cannot be guaranteed that the labels have been correctly<br />

assigned. On the other hand, if the labels are different, they can<br />

be merged by syntactically intersecting the dimension headers<br />

with a high threshold.<br />

4) Default<br />

Occasionally, the methods outlined in 1, 2, and 3 may fail<br />

to identify an appropriate label for the dimensions.<br />

Nevertheless, there may be a remaining list from the candidate<br />

list in heuristic 2 that has not been assigned as a label to any of<br />

the dimensions. The candidates derived from the dimension<br />

summary headers have the highest priority to be assigned as a<br />

label, because of the nature of this location in regards to the<br />

dimension. If a dimension summary header was not found, then<br />

the non-assigned candidate from the table title is assigned as<br />

the label without any processing.<br />

V. IDENTIFYING isA RELATIONSHIPS FROM A CANDIDATE<br />

SET<br />

In this section, the candidate set is constructed using the<br />

table title and the dimension summary header. One of the<br />

candidates in this set can be the dimension label. This paper<br />

introduces a special algorithm to reach this goal. The algorithm<br />

is run in three steps.<br />

In short, there should be a candidate set, i.e., { ,<br />

and the dimension headers; i.e., { ,, }. The problem is to<br />

select a , 1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!