Skip to main content
. 2020 Apr 23;8(4):e17642. doi: 10.2196/17642

Table 1.

Summary of the three strategies for improving keyword extraction performance.

Strategy Description Effect
Weight assignment Assign weight of 3, 1.2, and 0.8 to title words, nouns, and verbs, respectively, when performing keyword extraction. Nouns and title words will be more likely to be keywords, and verbs are less likely to be keywords.
Compound word identification Use several filter conditions to generate user-defined dictionary of compound words in educational materials for word segmentation. Compound words that meet filter conditions will be identified and are more likely to be the keywords than atom words.
Synonym elimination Remove shorter keywords with similar Chinese characters based on cosine similarity between their character compositions. Two or more keyword candidates with similar character composition will be merged into one keyword to avoid redundancy.