英语论文网

layerdistribution of semantic and syntactic features, which can show the interaction of semanticfeatures and syntactic features. The layer distribution is discussed in detail in the followingpart. The diagram of the layer distribution is shown in the following Figure 6.1. In the first layer, there are two features, a9 a1 respectively. They belong tosemantic features. Semantic features a9, namely, the zero co-occurrence between thetemporal meaning of over and the noun after it, and a1, the zero co-occurrence betweenthe spatial meaning of over and the noun after it, cover all of members of the objects. In the second layer, there are five semantic features and one syntactic feature. Thefive semantic features are a3, namely, the high-value of MI between the spatial meaning ofover and the noun after it, a1, namely, the zero co-occurrence between the spatial meaningof over and the noun after it, a6, namely, the relative low-value of MI between thenon-spatial meaning of over and the noun after it, a2, namely, the low-value of MIbetween the spatial meaning and the noun after it, a11, namely, the high-value of MIbetween the temporal meaning of over and the noun after it, and the syntactic feature isa14, namely, the abstract noun after over. It demonstrates that the second layer mainlyconsists of the semantic features.

…………

Conclusion

This paper constructs the WSD model by using formal concept analysis and bases themodel to explore the interactions between contextual features. The major findings arediscovered as follows:First of all, by the approach and the theory of formal concept analysis, the WSD modelof the English preposition over has been constructed. On the basis of the WSD model, thetwo testing sets are examined and the accuracy of disambiguation of over reaches 93%. Itindicates that the model constructed by formal concept analysis can disambiguate thesenses of over and achieve good results. But in order to avoid the problem of datasparseness, we put one training set and two testing sets all together and constitutes a largerdata. Another WSD model of over have been constructed in the same way. The accuracyof the model reaches 97.55%. It shows that this model can reach higher accuracy than theprevious model. It indicates that when the data set contains the more samples, the morepatterns would be extracted from the WSD model, so the model could get the higheraccuracy. The two models can lay the foundations for studying other semantic complicatedprepositions.
…………
Reference (omitted)

本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。