Code Generation from Natural Language with Less Prior and More Monolingual Data

Training datasets for semantic parsing are typically small due to the higher expertise required for annotation than most other NLP tasks. As a result, models for this application usually need additional prior knowledge to be built into the architecture or algorithm. The increased dependency on human...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2021-06
Hauptverfasser: Norouzi, Sajad, Tang, Keyi, Cao, Yanshuai
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Norouzi, Sajad
Tang, Keyi
Cao, Yanshuai
description Training datasets for semantic parsing are typically small due to the higher expertise required for annotation than most other NLP tasks. As a result, models for this application usually need additional prior knowledge to be built into the architecture or algorithm. The increased dependency on human experts hinders automation and raises the development and maintenance costs in practice. This work investigates whether a generic transformer-based seq2seq model can achieve competitive performance with minimal code-generation-specific inductive bias design. By exploiting a relatively sizeable monolingual corpus of the target programming language, which is cheap to mine from the web, we achieved 81.03% exact match accuracy on Django and 32.57 BLEU score on CoNaLa. Both are SOTA to the best of our knowledge. This positive evidence highlights a potentially easier path toward building accurate semantic parsers in practice.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2475214508</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2475214508</sourcerecordid><originalsourceid>FETCH-proquest_journals_24752145083</originalsourceid><addsrcrecordid>eNqNiksKwjAUAIMgWLR3eOC6kCaN7b7-FvWzcF8eNK0pMdF88PpW8ABuZhYzM5IwzvOsKhhbkNT7kVLKNiUTgifkUttOwkEa6TAoa6B39gFnDNGhhgbNEHGQ8FbhDo30Hq5OWQdoOjhZJycYq9X30rDFgCsy71F7mf68JOv97lYfs6ezryh9aEcbnZlSy4pSsLwQtOL_XR_AET3a</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2475214508</pqid></control><display><type>article</type><title>Code Generation from Natural Language with Less Prior and More Monolingual Data</title><source>Open Access: Freely Accessible Journals by multiple vendors</source><creator>Norouzi, Sajad ; Tang, Keyi ; Cao, Yanshuai</creator><creatorcontrib>Norouzi, Sajad ; Tang, Keyi ; Cao, Yanshuai</creatorcontrib><description>Training datasets for semantic parsing are typically small due to the higher expertise required for annotation than most other NLP tasks. As a result, models for this application usually need additional prior knowledge to be built into the architecture or algorithm. The increased dependency on human experts hinders automation and raises the development and maintenance costs in practice. This work investigates whether a generic transformer-based seq2seq model can achieve competitive performance with minimal code-generation-specific inductive bias design. By exploiting a relatively sizeable monolingual corpus of the target programming language, which is cheap to mine from the web, we achieved 81.03% exact match accuracy on Django and 32.57 BLEU score on CoNaLa. Both are SOTA to the best of our knowledge. This positive evidence highlights a potentially easier path toward building accurate semantic parsers in practice.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Algorithms ; Annotations ; Maintenance costs ; Natural language processing ; Parsers ; Programming languages ; Semantics</subject><ispartof>arXiv.org, 2021-06</ispartof><rights>2021. This work is published under http://creativecommons.org/licenses/by-nc-sa/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>780,784</link.rule.ids></links><search><creatorcontrib>Norouzi, Sajad</creatorcontrib><creatorcontrib>Tang, Keyi</creatorcontrib><creatorcontrib>Cao, Yanshuai</creatorcontrib><title>Code Generation from Natural Language with Less Prior and More Monolingual Data</title><title>arXiv.org</title><description>Training datasets for semantic parsing are typically small due to the higher expertise required for annotation than most other NLP tasks. As a result, models for this application usually need additional prior knowledge to be built into the architecture or algorithm. The increased dependency on human experts hinders automation and raises the development and maintenance costs in practice. This work investigates whether a generic transformer-based seq2seq model can achieve competitive performance with minimal code-generation-specific inductive bias design. By exploiting a relatively sizeable monolingual corpus of the target programming language, which is cheap to mine from the web, we achieved 81.03% exact match accuracy on Django and 32.57 BLEU score on CoNaLa. Both are SOTA to the best of our knowledge. This positive evidence highlights a potentially easier path toward building accurate semantic parsers in practice.</description><subject>Algorithms</subject><subject>Annotations</subject><subject>Maintenance costs</subject><subject>Natural language processing</subject><subject>Parsers</subject><subject>Programming languages</subject><subject>Semantics</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNiksKwjAUAIMgWLR3eOC6kCaN7b7-FvWzcF8eNK0pMdF88PpW8ABuZhYzM5IwzvOsKhhbkNT7kVLKNiUTgifkUttOwkEa6TAoa6B39gFnDNGhhgbNEHGQ8FbhDo30Hq5OWQdoOjhZJycYq9X30rDFgCsy71F7mf68JOv97lYfs6ezryh9aEcbnZlSy4pSsLwQtOL_XR_AET3a</recordid><startdate>20210610</startdate><enddate>20210610</enddate><creator>Norouzi, Sajad</creator><creator>Tang, Keyi</creator><creator>Cao, Yanshuai</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20210610</creationdate><title>Code Generation from Natural Language with Less Prior and More Monolingual Data</title><author>Norouzi, Sajad ; Tang, Keyi ; Cao, Yanshuai</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_24752145083</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Algorithms</topic><topic>Annotations</topic><topic>Maintenance costs</topic><topic>Natural language processing</topic><topic>Parsers</topic><topic>Programming languages</topic><topic>Semantics</topic><toplevel>online_resources</toplevel><creatorcontrib>Norouzi, Sajad</creatorcontrib><creatorcontrib>Tang, Keyi</creatorcontrib><creatorcontrib>Cao, Yanshuai</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</collection><collection>ProQuest Central Essentials</collection><collection>AUTh Library subscriptions: ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database (Proquest) (PQ_SDU_P3)</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Norouzi, Sajad</au><au>Tang, Keyi</au><au>Cao, Yanshuai</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Code Generation from Natural Language with Less Prior and More Monolingual Data</atitle><jtitle>arXiv.org</jtitle><date>2021-06-10</date><risdate>2021</risdate><eissn>2331-8422</eissn><abstract>Training datasets for semantic parsing are typically small due to the higher expertise required for annotation than most other NLP tasks. As a result, models for this application usually need additional prior knowledge to be built into the architecture or algorithm. The increased dependency on human experts hinders automation and raises the development and maintenance costs in practice. This work investigates whether a generic transformer-based seq2seq model can achieve competitive performance with minimal code-generation-specific inductive bias design. By exploiting a relatively sizeable monolingual corpus of the target programming language, which is cheap to mine from the web, we achieved 81.03% exact match accuracy on Django and 32.57 BLEU score on CoNaLa. Both are SOTA to the best of our knowledge. This positive evidence highlights a potentially easier path toward building accurate semantic parsers in practice.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2021-06
issn 2331-8422
language eng
recordid cdi_proquest_journals_2475214508
source Open Access: Freely Accessible Journals by multiple vendors
subjects Algorithms
Annotations
Maintenance costs
Natural language processing
Parsers
Programming languages
Semantics
title Code Generation from Natural Language with Less Prior and More Monolingual Data
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-07T15%3A43%3A55IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Code%20Generation%20from%20Natural%20Language%20with%20Less%20Prior%20and%20More%20Monolingual%20Data&rft.jtitle=arXiv.org&rft.au=Norouzi,%20Sajad&rft.date=2021-06-10&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2475214508%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2475214508&rft_id=info:pmid/&rfr_iscdi=true