Agriculture-related Snacks Sentiment Dictionary Construction for Cross-border E-commerce Platforms Based on SO-PMI Algorithm

In view of the problem that existing sentiment dictionaries cannot accurately capture and analyze the sentimental words of agriculture-related snacks in cross-border e-commerce platforms, a special sentiment dictionary is constructed by analyzing online reviews of agriculture-related snacks in cross-border e-commerce platforms. Firstly, Convolutional Neural Network (CNN) and Term Frequency (TF) are combined to calculate the sentiment seed words, and the candidate words are filtered by the text part-of-speech tagging. Then, the similarity between each candidate word and the seed emotion word was calculated by using SO-PMI algorithm to mark the polarity of the word. Finally, the formation of agriculture-related snacks sentiment dictionary. In order to verify the accuracy of the proposed dictionary, the constructed sentiment dictionary was used to classify the comments of agriculture-related snacks on different cross-border e-commerce platforms. The experimental results show that compared with other sentimental dictionaries, the accuracy rate, recall rate and F1 value of the three evaluation indicators have increased by more than 5.00 percentage points. The constructed sentiment dictionary of agricultural snacks has a better effect on emotion classification of agriculture-related snacks, which is convenient for cross-border e-commerce platform merchants to grasp consumers' sentimental tendency more accurately.