WikiCoder: Learning to Write Knowledge-Powered Code

Théo Matricon, Nathanaël Fijalkow, Gaëtan Margueritte
3/15/2023
cs.LGcs.PLcs.SE

No official code found. Best implementation: SynthesisLab/DeepSynth (22 stars, MIT, last commit 2026-01-03).

Abstract

We tackle the problem of automatic generation of computer programs from a few pairs of input-output examples. The starting point of this work is the observation that in many applications a solution program must use external knowledge not present in the examples: we call such programs knowledge-powered since they can refer to information collected from a knowledge graph such as Wikipedia. This paper makes a first step towards knowledge-powered program synthesis. We present WikiCoder, a system building upon state of the art machine-learned program synthesizers and integrating knowledge graphs. We evaluate it to show its wide applicability over different domains and discuss its limitations. WikiCoder solves tasks that no program synthesizers were able to solve before thanks to the use of knowledge graphs, while integrating with recent developments in the field to operate at scale.

View on arXivView PDF

Code Implementations(1)

SynthesisLab/DeepSynthCommunity100%
MIT

General purpose program synthetiser with existing Programming By Examples (PBE) pipeline

225PythonDec 6, 20219 months agoMIT
code-generationprogram-synthesis

Similar Papers

Assisted Knowledge Graph Authoring: Human-Supervised Knowledge Graph Construction from Natural Language

82% match1 repo92024

SAC-KG: Exploiting Large Language Models as Skilled Automatic Constructors for Domain Knowledge Graphs

78% match2024

Can humans teach machines to code?

78% match2024

Relevant Entity Selection: Knowledge Graph Bootstrapping via Zero-Shot Analogical Pruning

78% match1 repo92023

Context-Augmented Code Generation Using Programming Knowledge Graphs

77% match1 repo22026

Wikontic: Constructing Wikidata-Aligned, Ontology-Aware Knowledge Graphs with Large Language Models

77% match2025

Cite this paper

@article{matricon2023wikicoder,
  title  = {WikiCoder: Learning to Write Knowledge-Powered Code},
  author = {Théo Matricon and Nathanaël Fijalkow and Gaëtan Margueritte},
  year   = {2023},
  eprint = {2303.08574},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2303.08574v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2303.08574.svg)](https://codeofpaper.com/paper/2303.08574)

Discussion