IdBench: Evaluating Semantic Representations of Identifier Names in Source Code

Abstract

Identifier names convey useful information about the intended semantics of code. Name-based program analyses use this information, e.g., to detect bugs, to predict types, and to improve the readability of code. At the core of namebased analyses are semantic representations of identifiers, e.g., in the form of learned embeddings. The high-level goal of such a representation is to encode whether two identifiers, e.g., le n and s i z e , are semantically similar. Unfortunately, it is currently unclear to what extent semantic representations match the semantic relatedness and similarity perceived by developers. This paper presents IdBench, the first benchmark for evaluating semantic representations against a ground truth created from thousands of ratings by 500 software developers. We use IdBench to study state-of-the-art embedding techniques proposed for natural language, an embedding technique specifically designed for source code, and lexical string distance functions. Our results show that the effectiveness of semantic representations varies significantly and that the best available embeddings successfully represent semantic relatedness. On the downside, no existing technique provides a satisfactory representation of semantic similarities, among other reasons because identifiers with opposing meanings are incorrectly considered to be similar, which may lead to fatal mistakes, e.g., in a refactoring tool. Studying the strengths and weaknesses of the different techniques shows that they complement each other. As a first step toward exploiting this complementarity, we present an ensemble model that combines existing techniques and that clearly outperforms the best available semantic representation.Index Terms-source code, neural networks, embeddings, identifiers, benchmark I. In t r o d u c t io n Identifier names play an important role in writing, understanding, and maintaining high-quality source code [1]. Because they convey information about the meaning of variables, functions, classes, and other program elements, developers often rely on identifiers to understand code written by themselves and others. Beyond developers, various automated techniques analyze, use, and improve identifier names. For example, identifiers have been used to find programming errors [2 ]-[5], to mine specifications [6 ], to infer types [7], [8 ], to predict the name of a method [9], or to complete partial code using a learned language model [10]. Techniques for

On the Evaluation of NLP-based Models for Software Engineering

“…Relevance or Similarity: Several studies define relevance as to how relevant is the model's output to the reference text or code [2,5,11,16]. Others asked developers to rate the similarity, relatedness, and contextual or semantic similarity between outputs and reference texts [9,10,15].…”

Nalin: Learning from Runtime Behavior to Find Name-Value Inconsistencies in Jupyter Notebooks

“…To this end, the approach represents names and values as vectors that preserve their meaning. To represent identifier names, we build on learned token embeddings [13], which map each name into a vector while preserving the semantic similarities of names [54]. For example, the vector of probability will be close to the vectors of names probab and likelihood, because these names refer to similar concepts.…”

“…We build upon FastText [13], a neural word embedding known to represent the semantics of identifiers more accurately than other popular embeddings [54]. An additional key benefit of FastText is to avoid the out-of-vocabulary problem that other embeddings, e.g., Word2vec [36] suffer from, by splitting each token into n-grams and by computing a separate vector representation for each n-gram.…”

Naming Practices in Object-oriented Programming: An Empirical Study

“…As stated by Host and Ostvold (2007), even though naming is part of daily life for programmers, it entails a great deal of time and thought: names should convey to others the purpose of the code (Martin, 2008) and reflect the meaning of domain concepts (Marcus et al, 2004). Meaningful identifier names are key to bridging the gap between intention and implementation (Wainakh et al, 2021). Therefore, given that poorly chosen identifier names might hinder source code comprehension (Schankin et al, 2018), using meaningful identifier names is a recommended practice present in several coding style guides and conventions.…”