BEGIN:VCALENDAR VERSION:2.0 METHOD:PUBLISH PRODID:-//Missouri State University/Calendar of Events//EN CALSCALE:GREGORIAN X-WR-TIMEZONE:America/Chicago BEGIN:VTIMEZONE TZID:America/Chicago BEGIN:DAYLIGHT TZOFFSETFROM:-0600 TZOFFSETTO:-0500 DTSTART:20070311T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU TZNAME:CDT END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0500 TZOFFSETTO:-0600 DTSTART:20071104T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU TZNAME:CST END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:8575efba-a177-4ad9-a7da-b6cae33299d2.225353@calendar.missouristate.edu CREATED:20230209T190508Z LAST-MODIFIED:20230209T190508Z LOCATION: SUMMARY:PAMS Seminar: "Natural Language Processing for Accelerating Scient ific Breakthroughs" by John Dagdelen DESCRIPTION:John DagdelenUniversity of California Berkeley andLawrence Ber keley National Laboratory\n\n\nAbstract:The majority of all materials dat a is currently scattered across the text\, tables\, and figures of millio ns of scientific publications. In my talk\, I will discuss the work of ou r team at Lawrence Berkeley National Laboratory on the use of natural lan guage processing (NLP) to extract and discover scientific knowledge throu gh textual analysis of the abstracts of several million journal articles. With this data we are exploring new avenues for materials discovery and design such as how functional materials like thermoelectrics can be ident ified by using only unsupervised word embeddings for materials. To date\, we have used advanced techniques for named entity recognition to extract more than 100 million mentions of materials\, structures\, properties\, applications\, synthesis methods\, and characterization techniques from o ur database of over 3 million materials science abstracts. Our most recen t work utilizes GPT-3\, the same machine learning model behind OpenAI's C hatGPT\, for joint named entity recognition and relation extraction to ex tract complex hierarchical information from research articles. Finally\, I will also give an overview on how we are making all of this data freely available to the materials research community through our public-facing website matscholar.com and upcoming APIs.\n\n\nThis is a Zoom seminar. X-ALT-DESC;FMTTYPE=text/html:<html><head><title></title></head><body><p><b >John Dagdelen</b><br><b>University of California Berkeley and</b><br><b> Lawrence Berkeley National Laboratory</b></p>\n<p>Abstract:<br>The majori ty of all materials data is currently scattered across the text\, tables\ , and figures of millions of scientific publications. In my talk\, I will discuss the work of our team at Lawrence Berkeley National Laboratory on the use of natural language processing (NLP) to extract and discover sci entific knowledge through textual analysis of the abstracts of several mi llion journal articles. With this data we are exploring new avenues for m aterials discovery and design such as how functional materials like therm oelectrics can be identified by using only unsupervised word embeddings f or materials. To date\, we have used advanced techniques for named entity recognition to extract more than 100 million mentions of materials\, str uctures\, properties\, applications\, synthesis methods\, and characteriz ation techniques from our database of over 3 million materials science ab stracts. Our most recent work utilizes GPT-3\, the same machine learning model behind OpenAI's ChatGPT\, for joint named entity recognition and re lation extraction to extract complex hierarchical information from resear ch articles. Finally\, I will also give an overview on how we are making all of this data freely available to the materials research community thr ough our public-facing website matscholar.com and upcoming APIs.</p>\n<p> This is a Zoom seminar.</p></body></html> DTSTART;TZID=America/Chicago:20230223T160000 DTEND;TZID=America/Chicago:20230223T170000 SEQUENCE:0 URL:https://physics.missouristate.edu/seminars.htm CATEGORIES:Public,Alumni,Current Students,Faculty,Future Students,Staff END:VEVENT END:VCALENDAR