iai-group / cikm2019-table

Auto-completion for Data Cells in Relational Tables

This repository contains resources developed within the following paper:

S. Zhang and K. Balog. Auto-completion for Data Cells in Relational Tables. In: Proceedings of The 28th ACM International Conference on Information and Knowledge Management (CIKM ’19), Nov 2019.

Test collection

The table corpus is WikiTables, which comprises 1.6M tables extracted from Wikipedia. We proproceeed it and make it public downloadable here.

The data/queries.txt file contains the search queries.

The data/qrels.txt file contains the relevance assessments (in TREC qrels format).

Data

We utilize word2vec trained on Google news, and you can find it here. You can find the graph embeddings here.

Methods and results

The runfile/ folder contains the table rankings generated by the various methods (in TREC runfile format) (cf. Table 5 in the paper).

Source	Method	Runfile	Sources	used	Empty	excluded	Empty	included
			KB	TC	NDCG@5	NDCG@10	NDCG@5	NDCG@10
Single-source	KBLookupED		T		0.2635	0.2652	0.2780	0.2806
	InfoGather, top, UNI			T	0.4563	0.4710	0.4158	0.4302
	InfoGather, top, L2V			T	0.4868	0.4978	0.4413	0.4537
	TMatch, top, UNI			T	0.4744	0.4873	0.4297	0.4417
	TMatch, top, L2V			T	0.5046	0.5139	0.4531	0.4624
Multi-source	OTG		T	T	0.5856	0.6062	0.5185	0.5367
	CellAutoComplete (feat. I)		T	T	0.6641	0.6826	0.5766	0.5954
	CellAutoComplete (feat. I+II)		T	T	0.6844	0.7034	0.5905	0.6100
	CellAutoComplete (feat. I+II+III)		T	T	0.7570	0.7641	0.6716	0.6785

The evaluation scores are reported using trec_eval.

Citation

@inproceedings{Zhang:2019:ADC,
    author = {Zhang, Shuo and Balog, Krisztian},
    title = {Auto-completion for Data Cells in Relational Tables},
    booktitle = {Proceedings of the 28th ACM International Conference on Information and Knowledge Management (CIKM ’19)},
    year = {2019},
    pages = {},
}

Contact

If you have any questions, please contact Shuo Zhang at [email protected].