Pre-training method for sentiment analysis model, and electronic device

An electronic device, comprising:

• at least one processor;
• and
• a memory communicatively connected to the at least one processor;
• wherein,
• the memory stores instructions executable by the at least one processor, the instructions are executed by the at least one processor to enable the at least one processor to implement the method comprising:
• based on a given seed sentiment dictionary, performing sentimental knowledge detection on each training corpus in a training corpus set to determine a detection sentiment word and a detection word pair contained in the training corpus, wherein each detection word pair contains a comment point and a sentiment word;
• according to a preset mask processing rule, performing mask process on the detection sentiment word and the detection word pair in each training corpus to generate a masked corpus of the training corpus;
• performing encoding process on the masked corpus by using a preset encoder to generate feature vectors corresponding to the training corpus;
• performing decoding process on the feature vectors by using a preset decoder to determine a prediction sentiment word and a prediction word pair contained in the training corpus;
• and
• updating the preset encoder and the preset decoder according to a difference between the prediction sentiment word and the detection sentiment word, and a difference between the prediction word pair and the detection word pair.

A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are configured to cause a computer to implement the method comprising:

• based on a given seed sentiment dictionary, performing sentimental knowledge detection on each training corpus in a training corpus set to determine a detection sentiment word and a detection word pair contained in the training corpus, wherein each detection word pair contains a comment point and a sentiment word;
• according to a preset mask processing rule, performing mask process on the detection sentiment word and the detection word pair in each training corpus to generate a masked corpus of the training corpus;
• performing encoding process on the masked corpus by using a preset encoder to generate feature vectors corresponding to the training corpus;
• performing decoding process on the feature vectors by using a preset decoder to determine a prediction sentiment word and a prediction word pair contained in the training corpus;
• and
• updating the preset encoder and the preset decoder according to a difference between the prediction sentiment word and the detection sentiment word, and a difference between the prediction word pair and the detection word pair.