General commodity sequence representation learning method in recommendation system

The invention discloses a general commodity sequence representation learning method in a recommendation system. The general commodity sequence representation learning method comprises the following steps: S1, encoding associated texts of commodities by using a pre-training language model so as to learn transferable commodity representations; the method comprises the following steps: firstly, learning initial text representation by using a pre-training language model, and converting semantics of a text into a unified semantic space suitable for a recommendation task through a parameter whitening network and a hybrid expert enhanced adapter network; s2, further enhancing fusion and adaptation between data representations in different fields through a sequence-commodity comparison task and a sequence-sequence comparison task; and S3, considering two fine tuning settings, namely conclusion and transduction, according to whether the commodity label of the target domain is suitable for use or not. According to the