EmbeddingEntityRetrival

The source codes of the paper "Explore Entity Embedding Effectiveness in Entity Retrieval".

Main Baseline

There are codes for our main baselines:

• An Empirical Study of Learning to Rank for Entity Search. SIGIR 2016

Entity Baseline (ELR)

We utilize ELR feature from this paper:

• Exploiting Entity Linking in Queries for Entity Retrieval. ICTIR 2016

Run Your Models

• bash train_ca.sh
• bash train_svm.sh

Results

The entity retrieval results.

Model	SemSearch ES			ListSearch			INEX-LD			QALD-2			Total
	MAP	P@10	P@20	MAP	P@10	P@20	MAP	P@10	P@20	MAP	P@10	P@20	MAP	P@10	P@20
Baseline(CA)	0.3899	0.2908	0.2077	0.2334	0.3130	0.2378	0.1298	0.2900	0.2285	0.1998	0.1500	0.1196	0.2454	0.2540	0.1934
+ELR(CA)	0.3880	0.3023	0.2150	0.2443	0.3130	0.2422	0.1275	0.2920	0.2335	0.2074	0.1664	0.1282	0.2472	0.2544	0.1945
+TransE(CA)	0.4085	0.3023	0.2165	0.2507	0.3304	0.2543	0.1312	0.2860	0.2255	0.2270	0.1700	0.1371	0.2597	0.2639	0.1970
Baseline(SVM)	0.3895	0.3038	0.2169	0.2323	0.3078	0.2513	0.1350	0.2940	0.2345	0.2229	0.1529	0.1257	0.1925	0.2245	0.1798
+ELR(SVM)	0.3881	0.3046	0.2173	0.2390	0.3148	0.2530	0.1390	0.2980	0.2375	0.2197	0.1671	0.1286	0.2005	0.2307	0.1816
+TransE(SVM)	0.4061	0.3077	0.2196	0.2439	0.3252	0.2565	0.1392	0.2950	0.2365	0.2278	0.1629	0.1321	0.2054	0.2346	0.1881

Citation

Explore Entity Embedding Effectiveness in Entity Retrieval. Zhenghao Liu, Chenyan Xiong, Maosong Sun and Zhiyuan Liu. In Proceedings of CCL. Paper

Contact

If you have questions, suggestions and bug reports, please email [email protected].