Recognition of Bumblebee Behavior Based on AHP Model

Abstract

We propose a vespa mandarinia recognition model based on analytic hierarchy process to assess and classify reports of sightings. we construct the analytic hierarchy process (AHP) model to evaluate and classify reports. The four main indicators in the AHP model extracted from the data are image, position, note and detection time, and the corresponding weights of the four indicators are calculated. Then, we establish models based on the four indicators to determine the scoring criteria. For the notes part, we use Term Frequency-Inverse Document Frequency arithmetic to extract keywords and Long Short-Term Memory (LSTM) model to classify and score the notes. Based on the AHP model, scores of the report can be calculated. Next, we use the scores of the given reports to calculate the classification threshold and divide all the report into three status based on two thresholds: positive, negative and unverified. According to the corresponding formula, the two thresholds are calculated as 0.6276 and 0.3937 respectively.