diff options
| -rw-r--r-- | poster.odt | bin | 0 -> 43270 bytes | |||
| -rw-r--r-- | presentation1.md | 21 |
2 files changed, 21 insertions, 0 deletions
diff --git a/poster.odt b/poster.odt Binary files differnew file mode 100644 index 0000000..a94238b --- /dev/null +++ b/poster.odt diff --git a/presentation1.md b/presentation1.md new file mode 100644 index 0000000..269a9e8 --- /dev/null +++ b/presentation1.md @@ -0,0 +1,21 @@ +# Baseline Rankings +- On the left are the baseline rankings as they are given on the DBPedia-Entity +repository +- These baseline rankings are created by running the algorithms on a seemingly +arbitrary set of fields of the DBPedia Index +- Including some special fields that are noted on the right +- Noteworthy is the attribute section which includes a blacklist of fields that +are not used +- The blacklist appears to be created through educated but subjective guesses +- It is our goal to find out if this blacklist is ideal, or if it would be +better if it was changed + +# HillClimbing +- To achieve this, we intend to use Nordlys +- Nordlys includes the scripts that run the algorithms given a configuration +file, on the DBPedia Index. +- It is to our understanding that we can modify this configuration file by +adding and removing fields, and then run the algorithms +- To achieve a maximum, we intent to use a hill climbing algorithm with binary +weights; after all, we can only include or blacklist a field + |