At #acl2023nlp in Toronto, we present our work on directly retrieving facts from a knowledge graph without explicit entity & relation linking. Work with Jinheon Baek, Alham Fikri and Sung Ju Hwang.

Paper link: https://arxiv.org/pdf/2305.12416.pdf

#QuestionAnswering #LLMs #KnowledgeGraphs

📢📢 NEW BLOG POST FROM SPONSOR

We interviewed The QA Company, and asked them a few questions regarding their main products, their role in the Semantic Web Community, and about their vision.

Read the interview here: https://semantics-org.github.io/page/news?page=2023-07-01a

#semanticsconf #questionanswering

Updates on the #Qur'an #QuestionAnswering Shared tasks (Qur'an QA 2023) organized at WANLP 2023 Conference on December 7th, 2023 are available here:

https://sites.google.com/view/quran-qa-2023/home

You can find information on the two tasks and register until August 10th, 2023 on the website.

@msh
Not true. All the #benchmarks say otherwise. You have to look past the hyped #LLMs to the bread and butter BERT and BART models, but the trend is undeniable:

https://paperswithcode.com/area/natural-language-processing

#classification #retrieval #summarization #QuestionAnswering #translation #generation #NER #VQA

You name an NLP problem and there's an LLM that is now better at it than the average human. Not so 2 yrs ago. Times they are a change'n.
@futurebird

Everybody talks about #LLMs and #ChatGPT. A task they fail big is described in this article in Nature
@SciReports@twitter.com: https://www.nature.com/articles/s41598-023-33607-z

"The SciQA Scientific #QuestionAnswering Benchmark for Scholarly Knowledge" with
@OliverKarras@twitter.com,
@markusstocker@twitter.com,
@iamYaserJ@twitter.com et al.

If you think you can develop a QA system performing better than plain LLMs consider participating in Scholarly QALD
@iswc_conf@twitter.com

https://kgqa.github.io/scholarly-QALD-challenge/2023/

RT @mommi84@twitter.activitypub.actor

It is now possible to query RDF files using natural language with @gpt_index@twitter.activitypub.actor! Just plug your @OpenAI@twitter.activitypub.actor API key to enable explainable complex question answering powered by GPT-3. Supports RDF and RDFS vocabularies.

#RDF #KnowledgeGraph #QuestionAnswering #NLProc #XAI #AI #GPT3 #LLM

🐦🔗: https://twitter.com/mommi84/status/1631027355987656706

(2/) Key points:

* Human experts can detect ChatGPT-generated answers much more reliably than human amateurs (screenshot 1).

* Slightly more than half of the times human experts found ChatGPT-generated answers are helpful (screenshot 1).

* ChatGPT is more helpful in finance and psychology, and least in medical (screenshot 1).

* RoBERTa-based classifier can reliably detect ChatGPT-generated answers(96+ in F1; screenshot 2).

#AI #QuestionAnswering #NLP #NLProc #DeepLearning #Paper

(1/) Interesting paper comparing how distinguishable the answers from #ChatGPT and how helpful they are compared to those from humans when the same set of questions (screenshots) are asked.

Biyang Guo, Xin Zhang, Ziyuan Wang, Minqi Jiang, Jinran Nie, Yuxuan Ding, Jianwei Yue, and Yupeng Wu. 2023. “How Close Is ChatGPT to Human Experts? Comparison Corpus, Evaluation, and Detection.” arXiv [cs.CL]. arXiv. http://arxiv.org/abs/2301.07597

#AI #QuestionAnswering #NLP #NLProc #DeepLearning #Paper

There's plenty of discussion of using #llms such as #chatgpt for #informationretrieval and #questionanswering, but I don't think I've seen a quantitative comparison with more traditional models using a standard benchmark. Is there a paper somewhere covering this?

@astatide yeah there are good ways to use these models, but trying to generate when you want to search is definitely doing it wrong.

Generative models are drunk geniuses - they've seen all the knowledge in the world! But they're also drunk af. Like a modern oracle of Delphi, kinda!

But yes, search is: query understanding, then broad retrieval, then ranking, and finally returning _references_ which you can consult.

Even verging into #QuestionAnswering requires care, to avoid spewing nonsense.

Hello world! Here's my #introduction

I’m an #NLP #PhD student at #CMU and a Student Researcher at #Google, interested in #AI for {Language, Knowledge, Culture, Society}

Currently I work on #NLProc with applications to #ComputationalSocialScience, #ScienceOfScience, #QuestionAnswering, & #ScholarlyNLP.

I'm often inspired by #libraries, #wikis, #memex, etc -- long term I'd like to study & build interactive #KnowledgeAugmentation that supports discovery, agency, access, community, and trust.

Hello world! Here's my #introduction

I’m an #NLP #PhD student at #CMU and a Student Researcher at #Google, interested in #AI for {Language, Knowledge, Culture, Society}

Currently I work on #NLProc with applications to #ComputationalSocialScience, #ScienceOfScience, #QuestionAnswering, & #ScholarlyNLP.

I'm often inspired by #libraries, #wikis, #memex, etc -- long term I'd like to study & build interactive #KnowledgeAugmentation that supports discovery, agency, access, community, and trust.