Is collective intelligence helps more in polysemy tag optimazed algorithm than commonsense tool

Personalized information recommendation based on social tagging is now a hot domain for industry and academics, but lack of semantic information especially polysemy problem hamper its development. Based on an existing optimization algorithm, this paper puts forward a new polysemy tag optimization algorithm which adopts WordNet tool rather than tag co-occurrence, and gives out a new performance evaluation method to overcome the subjectivity problem of the traditional methods, which is focusing on the quality of recommendation, by means of the number of coincidence tags with target user. The experimental result shows that algorithm based on WordNet is better than tag co-occurrence, and the average recommendation quality value appear 10.2% higher than the latter.