Terminology spectrum analysis of natural-language chemical documents: term-like phrases retrieval routine

Table 9 Precision, Recall and F-measure estimated from the data obtained for 5 arbitrarily selected texts

Text no.	Number of terms retrieved by 2 experts	Number of term-like phrases retrieved by the system	Number of coincidences	Precision	Recall	F₁-measure
No. 1	164	221	135	0.61	0.82	0.70
No. 2	155	174	96	0.55	0.62	0.58
No. 3	170	172	113	0.66	0.66	0.66
No. 4	68	119	40	0.34	0.59	0.43
No. 5	125	215	106	0.50	0.85	0.63
*P, R and F values calculated for the entire 5 texts set:* Unique expert term-like phrases—655 Term-like n-grams—872 Coincidences—466				*0.53*	*0.71*	*0.61*

No. 1—Design, synthesis and catalysis of recoverable catalysts assembled in emulsion and…, C. Li et al. (2005)
No. 2—Understanding reaction pathways on model catalyst surfaces, F. Gao et al. (2007)
No. 3—Solid acid catalysts Based on H₃PW₁₂O₄₀ Heteropoly Acid: Acid and Catalytic Pr…, A.M. Alsalme et al. (2011)
No. 4—Advantages of using TOF–SIMS method in surface studies of heterogeneous…, M.I Szynkowska et al. (2005)
No. 5—ECS-Materials: synthesis and characterization of a new class of crystalline…, G. Bellussi et al. (2007)

ISSN: 1758-2946