Commodity evaluation analysis method based on multi-task deep learning
A technology of deep learning and analysis methods, applied in text database clustering/classification, special data processing applications, unstructured text data retrieval, etc., can solve the problems of inability to realize multi-task deep learning, poor accuracy, poor stability, etc. Achieve the effects of improving consumer experience, analyzing accuracy, improving generalization and robustness
Inactive Publication Date: 2019-10-01
HARBIN UNIV OF SCI & TECH
View PDF3 Cites 7 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] Aiming at the defects that the existing commodity evaluation and analysis methods cannot realize multi-task deep learning, poor stability, and poor accuracy, the present invention provides a commodity evaluation and analysis method that can realize multi-task deep learning, good stability, and high accuracy
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0041] to combine Figure 1-Figure 4 Describe this embodiment, in this embodiment, a kind of product evaluation analysis method based on multi-task deep learning involved in this embodiment, it comprises the following steps:
[0042] Step 1: Obtain the original text data set in the webpage, preprocess the text data set, and divide the text data set into training set and test set;
[0043] Step 2: After removing the stop words from the training set and the test set, use the word2vec word vector model to represent Chinese words as word vectors to form a word vector sequence;
[0044] Step 3: Input the output features of the word vector sequence as a model into the dual-channel LSTM network to share weights, use the sample pair-wise loss function to perform feature constraints in the middle layer of the neural network, and then learn through the gradient descent method;
[0045] Step 4: Use the softmax classification loss function at the top of the network to implement sentiment...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a commodity evaluation analysis method based on multi-task deep learning, and belongs to the field of natural language processing. The commodity evaluation analysis method provided by the invention can realize multi-task deep learning, and is good in stability and high in accuracy. The commodity evaluation analysis method comprises the following steps: preprocessing a textdata set, dividing the text data set into a training set and a test set, removing stop words, and representing Chinese words as word vectors by using a word2vec word vector model; inputting into a dual-channel LSTM (Long Short Term Memory) network to share a weight, carrying out feature constraint on a pair-wise loss function in a middle layer of a neural network by utilizing a sample, and learning through a gradient descent method; achieving emotion polarity analysis through a softmax classification loss function, learning feature distribution through a pair-wise loss function, and combiningthe softmax classification loss function and a pair of the pair-wise loss function for optimization. The commodity evaluation analysis method is mainly used for analyzing and processing commodity evaluation languages.
Description
technical field [0001] The invention belongs to the field of natural language processing under artificial intelligence, and specifically relates to a commodity evaluation analysis method based on multi-task deep learning. Background technique [0002] Commodity evaluation analysis is mainly based on sentiment analysis of reviews, and the existing indicators are comprehensive scoring systems to analyze commodities. The main analysis and research methods in the field of product evaluation and analysis are basically rule-based methods, traditional machine learning methods and deep learning methods, such as support vector machines, information entropy, conditional random fields, etc. Sentiment analysis in commodity evaluation first originated from text analysis based on grammatical rules. The method is relatively simple. Researchers with grammatical sensitivity need to establish a dictionary of sentiment analysis, and then divide the words expressing emotion into two categories....
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/35G06Q30/02
CPCG06F16/35G06F16/355G06Q30/0201
Inventor 谢金宝李瑞彤康守强王庆岩王玉静梁新涛
Owner HARBIN UNIV OF SCI & TECH