Weighted SCL model for adaptation of sentiment classification

In recent years, Structural Correspondence Learning (SCL) is regarded as one of the most promising techniques for transfer learning. The main idea behind SCL model is to identify correspondences among features from different domains by modeling their correlations with pivot features. However, SCL model treats each feature as well as each instance by an equivalent-weight strategy. From the perspective of feature, this strategy fails to overcome the adverse influence of high-frequency domain-specific (HFDS) features: they occupy a relative large portion of weight in classification model, while hardly carry corresponding sentiment information. From the other perspective, the equivalent-weight strategy of SCL model does not take into account the labels (“positive” or “negative”) of labeled instance and the labels of pivot features: positive pivot features tend to occur more frequently in positive instances and vice versa. To address the two issues effectively, we proposed a weighted SCL model (W-SCL), which weights the features as well as the instances. More specifically, W-SCL assigns a smaller weight to HFDS features and assigns a larger weight to instances with the same label as the involved pivot feature. The experimental results indicate that proposed W-SCL model could overcome the adverse influence of HFDS features, and leverage knowledge from labels of instances and pivot features.