ESBM: Entity Summarization BenchMark

• From: Gong Cheng <gcheng@nju.edu.cn>
• Date: Thu, 4 Jan 2018 01:02:59 +0800
• To: semantic-web@w3.org
• Message-Id: <142729C8-63A0-4AE7-B4EC-9DA7B5391D29@nju.edu.cn>

Dear colleagues,

We are happy to announce ESBM, a new benchmark for evaluating algorithms for entity summarization.

http://ws.nju.edu.cn/summarization/esbm/

ESBM consists of 140 entities selected from DBpedia and LinkedMDB. For each entity, we provide six ground-truth summaries independently created by different human experts. We also provide an evaluator written in Java, with which you can easily evaluate your own algorithm based on the benchmark.

We have used ESBM to evaluate a number of state-of-the-art algorithms. Their results are also included in our release, to be used as a baseline in your research.

Please do not hesitate to contact us for any questions or additional information about this benchmark.

Regards,

Gong Cheng
Nanjing University, China

Received on Wednesday, 3 January 2018 17:03:34 UTC