Exploiting Query’s Temporal Patterns for Query Autocompletion

Table 1

Performance of QAC models in terms of MRR on the AOL dataset for prefix consisting of 1–5 characters. The best results of baselines and of all models at each row are italic and bold, respectively.

	MPC-ALL	MPC-W					P-QAC
		2 days	4 days	7 days	14 days	28 days
1	0.1097	0.0868	0.1066	0.1113	0.1110	0.1109	0.1115
2	0.2012	0.1780	0.1894	0.1952	0.1985	0.2029	0.2059
3	0.3211	0.2653	0.2792	0.3005	0.3114	0.3195	0.3287
4	0.4155	0.3625	0.3918	0.4028	0.4126	0.4214	0.4356
5	0.4951	0.4370	0.4691	0.4800	0.4913	0.5028	0.5163