4 ms·

most of these knowledge graph papers seems to be about linking shallow content, likes citations etc, with all these strides in AI, GPT, etc, does anyone know o

by gryn 4y ago

most of these knowledge graph papers seems to be about linking shallow content, likes citations etc, with all these strides in AI, GPT, etc, does anyone know of papers about extracting into graph or other form of knowledge representations deeper semantics ? for example instead extracting what the paper is about, how it relates to other papers, nodes and edges linking to field specific terms/concepts mentioned, translation of paper specific terminology to commonly agreed terminology, why it's citing said references (is it an example of use of a technique ? a proof for a claim ? something that's being refuted, etc ..), where in the text it's being used. and other kinds of "deeper meaning". I've glanced at some papers that seemed to do something remotely close to this but it was restricted to doing NER for only medical/biology terms.

moonchrome 4y ago
I think this is the next step in AGI - feed and reference general knowledge graphs from/to LLM. Then link to expert systems such as SAT/facial recognition/physics model/medical model. I think LLMs will be capable of orchestrating between systems and maintaining a conscious narrative but knowledge graphs will solve the update and references problem.
stonerri 4y ago
The second layer is hard. I tried something in this space in mid-2018. Full text extraction and sentence segmentation tech was adequate, but extracting the discourse tree and building the graph was a bit of a struggle (trying to repurpose a collection of academic/open tools to get something useful). Never published or released the code. If interested, a few rabbit holes to explore (no affiliations): https://scite.ai https://scite.ai -> best option for citation mapping, but same issues you described above https://www.semanticscholar.org https://www.semanticscholar.org and AI2 -> the best group working on tooling in this space https://www.weave.bio https://www.weave.bio -> early startup trying to build this out The hardest challenge in my view is solving the intermediate representation issue. You have to establish a DSL/nomenclature that provides the range required to represent a complete scholastic discourse while also being computable.
- 21eleven 4y ago
  > The hardest challenge in my view is solving the intermediate representation issue. Right, you'd basically be writing an interpreter for English
  - nerpderp82 4y ago
    Have GPT translate down into a Controlled Natural Language. I tried having it translate to OWL, but it sucked. https://en.wikipedia.org/wiki/Controlled_natural_language https://en.wikipedia.org/wiki/Controlled_natural_language
    - uticus 4y ago
      “OWL”?
      - xyzzy3000 4y ago
        I assumed OWL='Web Ontology Language', which seems to make sense given that the topic is semantics - but OWL isn't a controlled vocabulary of a natural language, so I may be wrong.
        nerpderp82 4y ago
        You are correct, I just reached for the nearest machine readable knowledge format.
minxomat 4y ago
I'm building this, for (mostly) non-scientific non-fiction works (books, articles, news, etc.). Launching soon, with about 7,500 books indexed. Generally, what I found useful to build a graph between "topics" or entities was to use a HyDE[1] prompt to generate possible distinct definitions and then build a nearest-neighbor network from that. This successfully identifies related concepts in the truly abstract sense rather than literal entities. [1] https://summarity.com/hyde https://summarity.com/hyde
- bwb 4y ago
  I'd love to chat! I am doing a rough version of this at Shepherd.com and laying the ground work for more :) (ben@re-moveshepherd.com) I haven't uses Wikidata info yet, but hoping to expand to that in 3 or 4 months.
  - minxomat 4y ago
    Sure. Sent a ping.
- abhinavkulkarni 4y ago
  Hey @minxomat, HyDE sounds like an interesting approach. All dense retrieval approaches suffer from the problem you outlined in the blogpost. Have you looked at keyword-based or late-interaction models for retrieval such as ColBERTv2[1]? I find that late-interaction methods seem to offer best trade-off between semantic intelligence (precision) and retriving relevant documents (recall). [1] https://github.com/stanford-futuredata/ColBERT https://github.com/stanford-futuredata/ColBERT
beckman466 4y ago
not a paper but a project: http://libscie.org http://libscie.org
zozbot234 4y ago
You can mark reasons for paper citations using CiTO, the citation types ontology: https://sparontologies.github.io/cito/current/cito.html https://sparontologies.github.io/cito/current/cito.html Not sure if there's anything standard along the same lines for more general annotation of e.g. Web-derived resources.
jeromechoo 4y ago
Not a paper, but NLP extraction into knowledge graph representations do exist and are in use today. For example, here's a general purpose NLP model that links organization and people entities (among others) to each other based on factual relationships described in the text — https://demo.nl.diffbot.com https://demo.nl.diffbot.com This semantic extraction can be extrapolated to most any trainable context. A useful one I've worked with involved mapping supplier-partner relationships. A well built supply chain graph can identify every layer of risk in a single supply chain and provide the provenance to back it up. IMO, the biggest blocker to more mainstream use of Knowledge Graphs (even in the commercial world) is an actually intuitive interface for knowledge exploration. The real market innovation behind GPT isn't its 175 billion parameters, its the prompt interface that makes ChatGPT so universally accessible.
- dberenstein1957 4y ago
  Lovely, for me coreference resolution also is a huge issue. I created this package (https://github.com/Pandora-Intelligence/crosslingual-coreference https://github.com/Pandora-Intelligence/crosslingual-corefer...) which is currently the only viable solution to do coref resolution in Dutch and some other low-resource language. Also, if you want to scale, LLMs are going to prove to expensive, so eventually data need to be logged somewhere to create a fine-tuned model that can do sub-tasks and ideally do them better. What do you think?
  - jeromechoo 4y ago
    Nice! Yes, coreference resolution is surprisingly absent even in enterprise NLP. Depends on what you mean by "better". With more accuracy?

by gryn 4y ago

moonchrome 4y ago

I think this is the next step in AGI - feed and reference general knowledge graphs from/to LLM. Then link to expert systems such as SAT/facial recognition/physics model/medical model. I think LLMs will be capable of orchestrating between systems and maintaining a conscious narrative but knowledge graphs will solve the update and references problem.

stonerri 4y ago

The second layer is hard. I tried something in this space in mid-2018. Full text extraction and sentence segmentation tech was adequate, but extracting the discourse tree and building the graph was a bit of a struggle (trying to repurpose a collection of academic/open tools to get something useful). Never published or released the code. If interested, a few rabbit holes to explore (no affiliations): https://scite.ai https://scite.ai -> best option for citation mapping, but same issues you described above https://www.semanticscholar.org https://www.semanticscholar.org and AI2 -> the best group working on tooling in this space https://www.weave.bio https://www.weave.bio -> early startup trying to build this out The hardest challenge in my view is solving the intermediate representation issue. You have to establish a DSL/nomenclature that provides the range required to represent a complete scholastic discourse while also being computable.

21eleven 4y ago
> The hardest challenge in my view is solving the intermediate representation issue. Right, you'd basically be writing an interpreter for English
- nerpderp82 4y ago
  Have GPT translate down into a Controlled Natural Language. I tried having it translate to OWL, but it sucked. https://en.wikipedia.org/wiki/Controlled_natural_language https://en.wikipedia.org/wiki/Controlled_natural_language
  - uticus 4y ago
    “OWL”?
    - xyzzy3000 4y ago
      I assumed OWL='Web Ontology Language', which seems to make sense given that the topic is semantics - but OWL isn't a controlled vocabulary of a natural language, so I may be wrong.
      - nerpderp82 4y ago
        You are correct, I just reached for the nearest machine readable knowledge format.

minxomat 4y ago

I'm building this, for (mostly) non-scientific non-fiction works (books, articles, news, etc.). Launching soon, with about 7,500 books indexed. Generally, what I found useful to build a graph between "topics" or entities was to use a HyDE[1] prompt to generate possible distinct definitions and then build a nearest-neighbor network from that. This successfully identifies related concepts in the truly abstract sense rather than literal entities. [1] https://summarity.com/hyde https://summarity.com/hyde

bwb 4y ago
I'd love to chat! I am doing a rough version of this at Shepherd.com and laying the ground work for more :) (ben@re-moveshepherd.com) I haven't uses Wikidata info yet, but hoping to expand to that in 3 or 4 months.
- minxomat 4y ago
  Sure. Sent a ping.
abhinavkulkarni 4y ago
Hey @minxomat, HyDE sounds like an interesting approach. All dense retrieval approaches suffer from the problem you outlined in the blogpost. Have you looked at keyword-based or late-interaction models for retrieval such as ColBERTv2[1]? I find that late-interaction methods seem to offer best trade-off between semantic intelligence (precision) and retriving relevant documents (recall). [1] https://github.com/stanford-futuredata/ColBERT https://github.com/stanford-futuredata/ColBERT

beckman466 4y ago

not a paper but a project: http://libscie.org http://libscie.org

zozbot234 4y ago

You can mark reasons for paper citations using CiTO, the citation types ontology: https://sparontologies.github.io/cito/current/cito.html https://sparontologies.github.io/cito/current/cito.html Not sure if there's anything standard along the same lines for more general annotation of e.g. Web-derived resources.

jeromechoo 4y ago

Not a paper, but NLP extraction into knowledge graph representations do exist and are in use today. For example, here's a general purpose NLP model that links organization and people entities (among others) to each other based on factual relationships described in the text — https://demo.nl.diffbot.com https://demo.nl.diffbot.com This semantic extraction can be extrapolated to most any trainable context. A useful one I've worked with involved mapping supplier-partner relationships. A well built supply chain graph can identify every layer of risk in a single supply chain and provide the provenance to back it up. IMO, the biggest blocker to more mainstream use of Knowledge Graphs (even in the commercial world) is an actually intuitive interface for knowledge exploration. The real market innovation behind GPT isn't its 175 billion parameters, its the prompt interface that makes ChatGPT so universally accessible.

dberenstein1957 4y ago
Lovely, for me coreference resolution also is a huge issue. I created this package (https://github.com/Pandora-Intelligence/crosslingual-coreference https://github.com/Pandora-Intelligence/crosslingual-corefer...) which is currently the only viable solution to do coref resolution in Dutch and some other low-resource language. Also, if you want to scale, LLMs are going to prove to expensive, so eventually data need to be logged somewhere to create a fine-tuned model that can do sub-tasks and ideally do them better. What do you think?
- jeromechoo 4y ago
  Nice! Yes, coreference resolution is surprisingly absent even in enterprise NLP. Depends on what you mean by "better". With more accuracy?