Semisupervised Learning Based Opinion Summarization and Classification for Online Product Reviews

Table 3

Accuracy of opinionated sentence identification using mined product features: performance comparison of 3 strategies.

Product	Strategy 1			Strategy 2			Strategy 3
	Feature selection using Apriori with compactness and redundancy pruning			Feature selection using seed-set expansion approach			Feature selection using frequent multi-words with decomposition strategy
	Precision	Recall	𝐹-Measure	Precision	Recall	𝐹-Measure	Precision	Recall	𝐹-Measure
Tablet 1	0.8951	0.7183	0.7970	0.9174	0.6712	0.7752	0.9183	0.7894	0.8490
Tablet 2	0.7996	0.7477	0.7728	0.8593	0.7046	0.7743	0.8965	0.7647	0.8254
Tablet 3	0.8739	0.6994	0.7770	0.9453	0.6241	0.7518	0.9259	0.7462	0.8264
Tablet 4	0.8642	0.8190	0.8410	0.9287	0.5034	0.6529	0.9090	0.8163	0.8602
E-book Reader 1	0.8752	0.7842	0.8272	0.9011	0.6957	0.7852	0.9016	0.7857	0.8397
E-book Reader 2	0.9334	0.6905	0.7938	0.8945	0.6712	0.7669	0.9375	0.7627	0.8411
E-book Reader 3	0.9076	0.8013	0.8511	0.8291	0.7311	0.7770	0.9230	0.8135	0.8648
Average:	0.8784	0.7515	0.8100	0.8965	0.6573	0.7585	0.9160	0.7826	0.8441