A Systematic Literature Review of 10 years of Research on Program Synthesis and Natural Language Processing

A Systematic Literature Review of 10 years of Research on Program Synthesis and Natural Language Processing

연구 분야: Artificial Intelligence

학회: Programming and Computer Software

Program synthesis is the process of automatically generating software from a requirement specification. This paper presents a systematic literature review focused on program synthesis from specifications expressed in natural language. The research problem centers on the complexity of automatically generating accurate and robust code from high-level, ambiguous natural language descriptions – a barrier that limits the broader adoption of automatic code generation in software development. To address this issue, the study systematically examines research published between 2014 and 2024, focusing on works that explore various approaches to program synthesis from natural language inputs. The review follows a rigorous methodology, incorporating search strings tailored to capture relevant studies from five major data sources: IEEE, ACM, Springer, Elsevier, and MDPI. The selection process applied strict inclusion and exclusion criteria, resulting in a final set of 20 high-quality studies. The findings reveal significant advancements in the field, particularly in the integration of large language models (LLMs) with program synthesis techniques. The review also highlights the challenges and concludes by outlining key trends and proposing future research directions aimed at overcoming these challenges and expanding the applicability of program synthesis across various domains.