To help with API search, many automated API recommendation approaches have been proposed, and most of them leverage both Stack Overflow (SO) posts and API documentation to recommend APIs for a given natural language described programming task query. There are two orthogonal approaches for this task, i.e., information retrieval based and neural-based methods. Although these approaches have achieved remarkable performance, our observation shows that existing approaches can fail due to the following two reasons: 1) most information retrieval based approaches treat task queries as bags-of-words, which cannot capture the semantic-related sequential information, e.g., it is difficult for these approaches to distinguish the query \textit{convert int to string} from \textit{convert string to int}. 2) both the information retrieval based and the neural-based approaches are weak at distinguishing the semantic difference among lexically similar queries.

In this paper, we propose CLEAR, which leverages BERT sentence embedding and contrastive learning to tackle the above two issues of existing API recommendation approaches. Specifically, CLEAR embeds the whole sentence of posts with a BERT-based model rather than the bag-of-word model, which can preserve the semantic-related sequential information. In addition, CLEAR uses contrastive learning to learn the semantic representation of programming terminologies in different programming tasks. {Finally, CLEAR leverages the similarity between a given query and SO posts to recommend APIs for the query.} Our experiment results on three different test datasets confirm the effectiveness of CLEAR for both method-level and class-level API recommendation. Compared to the state-of-the-art API recommendation approaches, CLEAR improves the MAP by 25%-187% at method-level and 10%-100% at class-level.