[HN Gopher] How NASA Is Using Graph Technology and LLMs to Build...
___________________________________________________________________
How NASA Is Using Graph Technology and LLMs to Build a People
Knowledge Graph
Author : lexmo67
Score : 90 points
Date : 2025-04-27 16:22 UTC (6 hours ago)
(HTM) web link (memgraph.com)
(TXT) w3m dump (memgraph.com)
| rkwz wrote:
| Any idea which LLM they're using?
| mistrial9 wrote:
| NASA announced LLMs in early days (years ago) - it seemed like
| they wanted to understand their own document libraries! What else
| could be inferred here? mass layoffs plus "people substitutes" ?
| is there a more diplomatic way to see this?
| karamanolev wrote:
| The only connection between "they wanted to understand their
| own document libraries" and "mass layoffs" is potentially
| "increased efficiency leads to needing less people for the same
| job". If there's anything else, please let me know.
|
| And if it's that, then are you suggesting to _not_ implement a
| certain technological efficiency tool in order to keep (now
| clearly redundant) jobs? That has never worked long-term in the
| history of mankind, AFAIK.
| behnamoh wrote:
| > What made you choose memgraph? ... And then Memgraph showed me
| the cost. That kind of sold me for time for us to be able to do
| that.
|
| It's an ad post about memgraph.
| ctxc wrote:
| Yes, domain is memgraph and it seems to be a marketing case
| study.
| dpflan wrote:
| As an alternative to a pure graph db (e.g. here, memgraph), has
| anyone here used Apache's AGE graph-database extension for
| Postgresql? For making a knowledge graph that can live alongside
| SQL?
| dgllghr wrote:
| I believe AGE has unfortunately been defunded:
| https://github.com/apache/age/discussions/2150 It's a shame
| because it seemed like being able to query data across multiple
| paradigms would be really useful
| demaga wrote:
| > 27K nodes and 230K edges
|
| This is such an overkill for that kind of data. Even if they do
| plan to "scale up significantly", I doubt that they'll actually
| experience any benefit of graph db.
| mmooss wrote:
| Why do you say that?
| jerryseff wrote:
| Memgraph is laughably expensive - I honestly wonder what anyone
| actually uses it for outside of companies that just don't care
| about infra spend.
| mbuda wrote:
| DISCLAIMER: The co-founder and CTO of Memgraph here.
|
| To add more context, Memgraph Enterprise pricing is explained
| under https://memgraph.com/pricing: "Starting at $25,000 per
| year for 16 GB, Memgraph has an all-inclusive, simple pricing
| model that scales with your workload without restrictions. No
| charge for compute. No charge for replicas. No charge for
| algorithms. No Surprises.".
|
| In addition, Memgraph Community is free (standard BSL license,
| which turns into Apache2 4 years after release date, https://gi
| thub.com/memgraph/memgraph/blob/master/licenses/BS...), and it
| has many features that are usually considered enterprise
| (users, replication, not a single degradation in performance or
| scale, etc.).
|
| Please elaborate more about why the pricing seems expensive, or
| put it into the infra-cost perspective :pray:
| smarx007 wrote:
| I think on this site anything that's more expensive than free
| is considered expensive. Countless arguments have been had on
| Oracle vs Postgres, including lock-in. I think lock-in is
| more important to consider than license cost.
|
| To be fair, it is quite nice for the pricing to be
| transparent. And I think it's somewhat competitive w.r.t.
| Stardog, for example. The community version is less
| restricted than Ontotext, for example.
| smarx007 wrote:
| If you want a production-grade graph DBMS, you don't have that
| many OSS options that are reliable and well-supported.
|
| In the relational space, it took OSS options like Postgres many
| decades (and somehow paid-for person-years) to get to a place
| where enterprises seriously consider migrating off Oracle to
| it.
| jcgrillo wrote:
| Are there any? My experience so far with graph databases is a
| resounding failure.
| XorNot wrote:
| In OSS or generally?
| timewizard wrote:
| > It's [sql] just not built for the complex relationships that
| exist in a massive organizations like NASA.
|
| This is an absurd claim.
|
| > Extracted Skills from Team Resumes
|
| > Extracted Skills
|
| > Subject Matter Experts Finder Question: designed to identify
| employees with expertise in specific domains or mission-critical
| capabilities.
|
| I can't think of anything that screams "incompetent management"
| more than this. So, to find a subject matter expert, you're going
| to "extract skills" and "extract resumes" to answer abstract
| questions about your staff... without ever once.. just _talking_
| to your staff?
|
| What a cold and bizarre future these people think we want to live
| in.
|
| Meanwhile can we use technology to improve the level of
| connectivity I have and experience as an employee? Can you please
| stop asking LLMs to "extract" things about me into goofy
| automated pipelines? If you want skilled workers then you need to
| demonstrate skilled management. This is all the exact opposite of
| that.
| mbuda wrote:
| This is like saying: "Here we have a rocket, but let's keep
| trying to go to the moon by bike." xD
|
| What's wrong with attempting to better understand a given
| organization using LLMs or any other tech? Ofc, great managers
| will try as hard as possible to talk face to face as much as
| possible.
| timewizard wrote:
| > This is like saying:
|
| I highly doubt the difference between current staff
| management and adding this thin layer is equivalent to
| difference between a bike and a rocket. It's more like saying
| "we get to the moon just fine, but if we strap this extra
| booster on, we will get there 2% faster than before but with
| all kinds of additional risks to the payload!"
|
| > What's wrong with attempting to better understand a given
| organization
|
| You can alienate your employees and lose your skill base as a
| result. I'd like to be evaluated based on upon my work and
| dedication, not what some LLM thinks it sees in my resume.
| I've worked for my current company for 17 years. My resume
| contains none of that work or any skills gained in that time.
|
| I also like to take on new challenges and learn new skills.
| The LLMs "extractions" cannot see this or attend to it.
|
| > Ofc, great managers will try as hard as possible to talk
| face to face as much as possible.
|
| That's not the problem being discussed here. The question is
| "can we use technology to make better organizational
| decisions particularly when it comes to the efficient use of
| human resources." If I have a bad boss, I'm going to quit,
| and you'll never even have this opportunity. If I have a good
| boss, and you interfere with his decisions using LLM driven
| logic, I'm going to quit, and you're never going to get the
| benefit of that labor anyways.
| throwaway0123_5 wrote:
| It doesn't seem like it is intended for a manager to use to get
| information about their direct reports, but rather org-wide
| information?
|
| NASA apparently has ~18k employees, it seems like it might be
| useful to be able to query "Who at NASA has X, Y, Z skills that
| can help us with this project." Then you can speak to some of
| those people face-to-face. It won't be perfect but certainly
| sounds like a useful tool in principle.
| browningstreet wrote:
| Every large consulting org does this. It's a market for
| enterprise and home-grown solutions.
| soco wrote:
| My consulting org has 3 (three) right now running in parallel
| and had another few over the last decade. Yet people still
| send emails around asking for support. And as a consultant
| every one is mandatory and every one is a bad joke in its own
| way. I don't think the market is anything else than a huge
| scam. Is this tool here better? If yes, hats off.
| inerte wrote:
| I know it's a marketing case study, but:
|
| > Ever wondered how NASA identifies its top experts, forms high-
| performing teams, and plans for the skills of tomorrow?
|
| Here's another resource on that
| https://appel.nasa.gov/2010/02/18/aa_2-7_f_nasa_teams-html/ the
| book "How NASA Builds Teams: Mission Critical Soft Skills for
| Scientists, Engineers, and Project Teams"
| cebert wrote:
| I think I found a place Dodge can save some money. Memgraph
| pricing is ridiculous.
| patcon wrote:
| Even paying a college grad to babysit a server costs more than
| their yearly rate. I assume you're speaking as someone who
| loves to host everything for themselves, but the logic is
| surely different in enterprise/government, no?
| jandrewrogers wrote:
| > The current graph has about 27K nodes and 230K edges
|
| That is _tiny_ even by historical standards. I was expecting
| there to be some type of technology here. Why is this
| interesting?
| smarx007 wrote:
| > "To make sure everyone understands that, I prefer label
| property graphs over RDF."
|
| I have two major issues with virtually all graph DBMSs that are
| not RDF/SPARQL-based:
|
| 1) They do not allow structure-preserving querying. That is, I
| query a graph and want the results to be a smaller graph. This is
| trivial in SQL, you just 'SELECT * FROM x WHERE ...' and the
| result set you get is tabular just like the table x. In SPARQL,
| there are a CONSTRUCT/DESCRIBE queries that do just that - give
| you the results as a graph.
|
| 2) They don't use any (internationally recognized) standard to
| represent graph data. RDF is the only such format known to me
| (ignore all the semantic web stuff associated with it and just
| consider the format).
|
| 230k edges is peanuts for a graph db. It's like when the number
| of rows times columns in your SQL DB is 230k. NASA could
| (should?) have just used Oxigraph, RDF4J, or Jena. Stardog and
| Ontotext are the paid options. However, it is quite nice to see
| more interest in graph-based DBMSs in general!
|
| > "Which employees have cross-disciplinary expertise in AI/ML?"
|
| Regarding the study itself, I did not understand who is the
| target user of this. I would rather be more interested in the
| Lessons Learned 2.0 study (I understand it was attempted once
| before [1]). I don't think the study at hand would be able to
| correctly answer questions about expertise.
|
| On the technical side, as far as I understand, the cosine
| similarity was computed per triplet? In that case, I could see
| how pgvector could be used for this. Relevance expansion is the
| only thing in the article that made me think that it would be
| cool if it works well. But I could see how in a combo of a
| regular RDF DBMS + pgvector, one could first do a cosine
| similarity query via pgvector and then compute an (S)CBD [2] of
| the subject (the from node) of the triplet.
|
| [1]: https://youtu.be/QEBVoultYJg?t=1653
|
| [2]: https://patterns.dataincubator.org/book/bounded-
| description....
| gitroom wrote:
| Man, love seeing pushback on automated skill matchingsometimes
| feels like tech folks keep inventing new tools just to dodge
| actual conversations. Ever wonder if all this automation just
| makes things colder instead of smarter?
| citizenpaul wrote:
| My experience with tools like this is that they have only one
| single outcome. Piling work onto the most talented or
| desperate(ie need money or visa) people until they leave the
| org/company. Eventually leading to total skill erosion and a very
| low average skill/productivity across the company as people leave
| or hide their abilities.
|
| Why? because there is never a reward attached. Oh you want to
| make me the AI resource for the agency but not remove former
| duties or increase my pay? Ummmm no thanks. Also things tend to
| happen in waves ie "AI" so everyone needs a lot from a very few
| people at the same time. No one ever asks how those people can be
| empowered. Just how can we put the screws to them so they work
| harder.
|
| HR and Mgmt can f-off with their "skill resource bank" or
| whatever nonsense they call it this year. My skills are what I
| was hired for on the job description. If you want to discuss a
| new position or higher pay for different skills I'm very happy to
| talk about how I can work with the org to make that happen. Thats
| never the case though.
___________________________________________________________________
(page generated 2025-04-27 23:00 UTC)