In addition, this paper proposes an update method for calculating the weight of feature words after synonym merging to obtain a more accurate vector representing of the text for classifier processing.
The parameter vector of the ({text{Child}} = left[ {{text{parC}}_{ 1},cdots cdots, {text{parC}}_{ 2 5} } right]) generated by the mutation is given by {text{parC}}_{text{i}} = {text{parP1}}_{text{i}} + {text{Coe}}_{text{i}}.
The parameter vector of the ({text{Child}} = left[ {{text{parC}}_{ 1},cdots cdots, {text{parC}}_{ 2 5} } right]) generated by the crossover is given by {text{parC}}_{text{i}} = {text{parP1}}_{text{i}} times {text{Coe}}_{text{i}} + {text{parP2}}_{text{i}} times left| {{text{Coe}}_{text{i}} - 1} right|.
In reference [104], CNN convolves and abstracts word vectors of the original text with filters of a certain length, and thus previous pure word vector become convolved abstract sequences.
In [6, 7], the authors utilize a vector of cipher text to mask the bidding prices, and guarantee that the maximum value, randomizing the bids, and charging the bidders can be figured out.
Then, in the next step, an improved TF-IDF method is used to calculate the feature weights for each word to generate the feature vector of each text.
The amount of classification included in training sets is exactly the dimensionality of CI subspace, which usually is smaller than that of the text vector space, so dimensionality reduction of vector space is achieved.
As described in (Soucy & Mineau [22]), supervised weighting methods such as Gain Ratio based on statistical confidence intervals could be used to improve the quality of the text vectors and, as a consequence, the clustering results.
This is the subcloning vector called pBACLinkSpAB in the rest of the text.
In [6 10], homomorphic encryption [11 13] is employed to hid each bidder's bidding values with a vector of cipher texts, and ensures the auctioneer to figure out the maximum value, and charge the bidders securely.
