Text Classification Using Novel Term Weighting Scheme-Based Improved TF-IDF for Internet Media Reports

With the rapid development of internet technology, a large amount text data can be obtained. The classification (TC) technology plays very important role in processing massive data, but accuracy is directly affected by performance term weighting TC. Due to original design information retrieval (IR), frequency-inverse document frequency (TF-IDF) not effective enough for TC, especially with unbalanced distributions media reports. Therefore, variance between DF value particular and average all DFs DF ¯ , namely, (ADF), proposed enhance ability distribution. Then, normal TF-IDF modified ADF collection four different ways, TF-IADF, TF-IADF+, TF-IADFnorm, TF-IADF+norm. As result, an model established TC task A series simulations have been carried out evaluate methods. Compared on state-of-the-art algorithms, effectiveness feasibility methods are confirmed simulation results.