Chinese semantic matching method based on twinborn interaction and fine tuning representation
The invention discloses a Chinese semantic matching method based on twinborn interaction and fine tuning representation, which comprises the following steps of: firstly, completing vector initialization of a text by using a RoBERTa-WWM-EXT pre-training model, and constructing a twinborn structure em...
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Patent |
| Sprache: | chi ; eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | LI BAOLIAN QIANG BAOHUA XI GUANGYONG CHEN JINYONG WANG YUFENG |
| description | The invention discloses a Chinese semantic matching method based on twinborn interaction and fine tuning representation, which comprises the following steps of: firstly, completing vector initialization of a text by using a RoBERTa-WWM-EXT pre-training model, and constructing a twinborn structure embedded with a soft alignment attention mechanism (SA-attention) and a BiLSTM training layer aiming at an initial feature vector so as to enhance the semantic interaction between sentence pairs; secondly, two texts to be matched are connected to be connected into a RoBERTa-WWM-EXT pre-training model for vectorization, and a connected vectorization result is input into an LSTM-BiLSTM network layer for enhancement training so as to strengthen the upper and lower semantic relation in a sentence; and then a training model capable of finely tuning RoBERTa-WWM-EXT initial vectors is built to generate text vectors subjected to tag supervision fine tuning, so that the expression strength of the vectors on the semantic relat |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_CN115345175A</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>CN115345175A</sourcerecordid><originalsourceid>FETCH-epo_espacenet_CN115345175A3</originalsourceid><addsrcrecordid>eNqNjDEKAjEQRdNYiHqH8QAWYQ3WEhQrK1tZZpNZN7CZhGTE65sFD2D14f3_31o97RSYKkGliCzBQURxjb0gkkzJw4CVPCQG-QQeUmEILFTQSWgQ2cPYDCBvXk6Fcmk6FlzqrVqNOFfa_XKj9tfLw94OlFNPNaMjJuntXWvTHY0-mXP3z-YLNvA8zw</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Chinese semantic matching method based on twinborn interaction and fine tuning representation</title><source>esp@cenet</source><creator>LI BAOLIAN ; QIANG BAOHUA ; XI GUANGYONG ; CHEN JINYONG ; WANG YUFENG</creator><creatorcontrib>LI BAOLIAN ; QIANG BAOHUA ; XI GUANGYONG ; CHEN JINYONG ; WANG YUFENG</creatorcontrib><description>The invention discloses a Chinese semantic matching method based on twinborn interaction and fine tuning representation, which comprises the following steps of: firstly, completing vector initialization of a text by using a RoBERTa-WWM-EXT pre-training model, and constructing a twinborn structure embedded with a soft alignment attention mechanism (SA-attention) and a BiLSTM training layer aiming at an initial feature vector so as to enhance the semantic interaction between sentence pairs; secondly, two texts to be matched are connected to be connected into a RoBERTa-WWM-EXT pre-training model for vectorization, and a connected vectorization result is input into an LSTM-BiLSTM network layer for enhancement training so as to strengthen the upper and lower semantic relation in a sentence; and then a training model capable of finely tuning RoBERTa-WWM-EXT initial vectors is built to generate text vectors subjected to tag supervision fine tuning, so that the expression strength of the vectors on the semantic relat</description><language>chi ; eng</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; ELECTRIC DIGITAL DATA PROCESSING ; PHYSICS</subject><creationdate>2022</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20221115&DB=EPODOC&CC=CN&NR=115345175A$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25542,76289</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20221115&DB=EPODOC&CC=CN&NR=115345175A$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>LI BAOLIAN</creatorcontrib><creatorcontrib>QIANG BAOHUA</creatorcontrib><creatorcontrib>XI GUANGYONG</creatorcontrib><creatorcontrib>CHEN JINYONG</creatorcontrib><creatorcontrib>WANG YUFENG</creatorcontrib><title>Chinese semantic matching method based on twinborn interaction and fine tuning representation</title><description>The invention discloses a Chinese semantic matching method based on twinborn interaction and fine tuning representation, which comprises the following steps of: firstly, completing vector initialization of a text by using a RoBERTa-WWM-EXT pre-training model, and constructing a twinborn structure embedded with a soft alignment attention mechanism (SA-attention) and a BiLSTM training layer aiming at an initial feature vector so as to enhance the semantic interaction between sentence pairs; secondly, two texts to be matched are connected to be connected into a RoBERTa-WWM-EXT pre-training model for vectorization, and a connected vectorization result is input into an LSTM-BiLSTM network layer for enhancement training so as to strengthen the upper and lower semantic relation in a sentence; and then a training model capable of finely tuning RoBERTa-WWM-EXT initial vectors is built to generate text vectors subjected to tag supervision fine tuning, so that the expression strength of the vectors on the semantic relat</description><subject>CALCULATING</subject><subject>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>ELECTRIC DIGITAL DATA PROCESSING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2022</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNqNjDEKAjEQRdNYiHqH8QAWYQ3WEhQrK1tZZpNZN7CZhGTE65sFD2D14f3_31o97RSYKkGliCzBQURxjb0gkkzJw4CVPCQG-QQeUmEILFTQSWgQ2cPYDCBvXk6Fcmk6FlzqrVqNOFfa_XKj9tfLw94OlFNPNaMjJuntXWvTHY0-mXP3z-YLNvA8zw</recordid><startdate>20221115</startdate><enddate>20221115</enddate><creator>LI BAOLIAN</creator><creator>QIANG BAOHUA</creator><creator>XI GUANGYONG</creator><creator>CHEN JINYONG</creator><creator>WANG YUFENG</creator><scope>EVB</scope></search><sort><creationdate>20221115</creationdate><title>Chinese semantic matching method based on twinborn interaction and fine tuning representation</title><author>LI BAOLIAN ; QIANG BAOHUA ; XI GUANGYONG ; CHEN JINYONG ; WANG YUFENG</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_CN115345175A3</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>chi ; eng</language><creationdate>2022</creationdate><topic>CALCULATING</topic><topic>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>ELECTRIC DIGITAL DATA PROCESSING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>LI BAOLIAN</creatorcontrib><creatorcontrib>QIANG BAOHUA</creatorcontrib><creatorcontrib>XI GUANGYONG</creatorcontrib><creatorcontrib>CHEN JINYONG</creatorcontrib><creatorcontrib>WANG YUFENG</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>LI BAOLIAN</au><au>QIANG BAOHUA</au><au>XI GUANGYONG</au><au>CHEN JINYONG</au><au>WANG YUFENG</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Chinese semantic matching method based on twinborn interaction and fine tuning representation</title><date>2022-11-15</date><risdate>2022</risdate><abstract>The invention discloses a Chinese semantic matching method based on twinborn interaction and fine tuning representation, which comprises the following steps of: firstly, completing vector initialization of a text by using a RoBERTa-WWM-EXT pre-training model, and constructing a twinborn structure embedded with a soft alignment attention mechanism (SA-attention) and a BiLSTM training layer aiming at an initial feature vector so as to enhance the semantic interaction between sentence pairs; secondly, two texts to be matched are connected to be connected into a RoBERTa-WWM-EXT pre-training model for vectorization, and a connected vectorization result is input into an LSTM-BiLSTM network layer for enhancement training so as to strengthen the upper and lower semantic relation in a sentence; and then a training model capable of finely tuning RoBERTa-WWM-EXT initial vectors is built to generate text vectors subjected to tag supervision fine tuning, so that the expression strength of the vectors on the semantic relat</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | chi ; eng |
| recordid | cdi_epo_espacenet_CN115345175A |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING ELECTRIC DIGITAL DATA PROCESSING PHYSICS |
| title | Chinese semantic matching method based on twinborn interaction and fine tuning representation |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T16%3A29%3A21IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=LI%20BAOLIAN&rft.date=2022-11-15&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3ECN115345175A%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |