HackerReposA FIELD GUIDE TO OPEN SOURCE
ProSponsor the shelf♧ 0
← Back to top all-time
stassa
louise
Prolog2019–2026
stassa / louise

louise

Polynomial-time Meta-Interpretive Learning

6Distinct people
25Mentions
1Explicit recommenders
137GitHub stars
REPOSITORY AT A GLANCE

What to know before exploring

What it is

Polynomial-time Meta-Interpretive Learning

GitHub

Where it is used

No documented product use has been verified yet.

What people are saying

Hey, if you're interested in higher-order logic programming then you might want to have a look at Meta-Interpretive Learning (MIL). It's a new approach to Inductive Logic…

YeGoblynQueenne used a recommendation phrase. One indexed comment matches. This is a phrase match, not a verdict.

No classified warning is attached to this page. Read the discussions before treating silence as approval.

Also in these discussions

No other repository is repeatedly linked in the same discussions.

This week on Hacker News

Nov 9–Nov 16 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/16/2026, 3:16:21 AM UTC.

No direct repository links were found in the indexed HN discussions over the past seven days. Older discussions are available below.

Mentions over time

HN mentions by year25 dated mentions

Observed mentions: 2019-08-04 07:50:07 UTC to 2026-02-24 18:03:33 UTC.

02462019-01-01 00:00:00 UTC to 2020-01-01 00:00:00 UTC (exclusive): 1 mentions; 0 submissions, 1 comments20192020-01-01 00:00:00 UTC to 2021-01-01 00:00:00 UTC (exclusive): 3 mentions; 1 submissions, 2 comments2021-01-01 00:00:00 UTC to 2022-01-01 00:00:00 UTC (exclusive): 4 mentions; 0 submissions, 4 comments2022-01-01 00:00:00 UTC to 2023-01-01 00:00:00 UTC (exclusive): 3 mentions; 0 submissions, 3 comments20222023-01-01 00:00:00 UTC to 2024-01-01 00:00:00 UTC (exclusive): 6 mentions; 0 submissions, 6 comments2024-01-01 00:00:00 UTC to 2025-01-01 00:00:00 UTC (exclusive): 4 mentions; 0 submissions, 4 comments2025-01-01 00:00:00 UTC to 2026-01-01 00:00:00 UTC (exclusive): 2 mentions; 0 submissions, 2 comments2026-01-01 00:00:00 UTC to 2027-01-01 00:00:00 UTC (exclusive): 2 mentions; 0 submissions, 2 comments. Incomplete calendar period.2026*

* An incomplete calendar period is marked with an outlined bar. Its count is not directly comparable with a full period.

Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.

View mention counts and date ranges
UTC ranges include their start and exclude their end. Select a period to filter the source evidence.
PeriodStart (UTC)End (exclusive, UTC)MentionsSubmissionsComments
20192019-01-01 00:00:00 UTC2020-01-01 00:00:00 UTC101
20202020-01-01 00:00:00 UTC2021-01-01 00:00:00 UTC312
20212021-01-01 00:00:00 UTC2022-01-01 00:00:00 UTC404
20222022-01-01 00:00:00 UTC2023-01-01 00:00:00 UTC303
20232023-01-01 00:00:00 UTC2024-01-01 00:00:00 UTC606
20242024-01-01 00:00:00 UTC2025-01-01 00:00:00 UTC404
20252025-01-01 00:00:00 UTC2026-01-01 00:00:00 UTC202
2026 (partial)2026-01-01 00:00:00 UTC2027-01-01 00:00:00 UTC202

What Hacker News says

4 sources
My guess is the end result of all this "AI" assisted code-generation is that it will have the same impact on the software engineering industry as spreadsheets had on accounting. I also believe that this AI-powered stuff is a bit of a "two-steps forward, one step back" situation and the real innovation will begin when ideas from tools like Louise [1] are integrated into the approach taken in Codex. When VisiCalc was released departments of 30 accountants were reduced to 5 accountants because of the improvement for individual worker efficiency, however accounting itself remains largely unchanged and accountants are still a respected profession who perform important functions. There's plenty of programming problems in the world that simply aren't being solved because we haven't figured out how to reduce the burden of producing the software;
Read the rest
code generation will simply increase the output of an individual software developer. The same forces behind "no-code" are at work here. In fact I see a future where these two solutions intermingle: where "no-code" becomes synonymous with prompt-driven development. As we all know, however, these solutions will only take you so far -- and essentially only allow you to express problems in domains that are already well-solved. We're just expressing a higher level of program abstraction; programs that generate programs. This is a good thing and it is not a threat to the existence of our industry. Even in Star Trek they still have engineers who fix their computers... [1] - https://github.com/stassa/louise
In How good is Codex?
>> 3) all predicates work in all-modes, all the time I think this is the feature that is the most sorely missed in Prolog programming and as you say it is possible to have it, but only if one is prepared to jump through no end of hoops. Sometimes it's just not possbile with any reasonable amount of effort and any sensible amount of coding. This feature is also the one most characteristic of relations, rather than functions, or the demi-functions that Prolog predicate definitions end up as (i.e. with more or less standard inputs and outputs, a.k.a. modes). I think I detect a hint of frustration in your comment, perhaps after one too many encounter with recalcitrant Prolog programmers, unwilling to give up their convenient hacks :) Well, these days I can wear another hat, that of the ILP researcher, and there is an interesting use of modes
Read the rest
in ILP: they are often used to reduce the size of the hypothesis search space. Because, of course, the hypothesis space for the program append(+,+,-) is much smaller than the hypothesis space for the program append(?,?,?)! Aleph, probably the best-known ILP system of all time (and its ancestor, Progol) has a special syntax of "mode declarations" to determine input and output modes, for this reason. Interestingly, the more recent approach (attention: I'm about to talk about my PhD work :) of Meta-Interpretive Learning (MIL), does away with specific modes and Aleph and Progol's (and many other systems') mode declarations. Instead, MIL encodes inductive bias by means of second-order datalog clauses, the "metarules" (e.g. ∃P,Q,R,X ∀x,y,z: P(x,y) ← Q(x,z), R(z,y,X) is a metarule with second-order variables P,Q,R existentially quantified over the set of predicates, first-order variable X existentially quantified over the set of constants and first-order variables x,y,z universally quantified over constants). Metarules do not specify input and output modes -they are, after all, datalog. This would normally blow up the hypothesis space to unsearchable proportions, but this is managed in various ways- and an actual search of the hypothesis space (i.e. the set of sets of clauses) can be avoided alltogether, for a polynomial time learning approach. But I digress. What I mean to say is that this theoretically mode-free learning has so far only been implementedin Prolog [1][2][3] or ASP [4]. ASP is purely declarative, but incomplete in its own way (because the Herbrand base must be fully ground, which is impossible to do for infinite Herbrand bases) and Prolog has- the mode problem we discussed. So while we have a powerful learning approach, with a nice theoretical basis, in practice we implement it in a less than satisfying manner, always. And, I dare say I've seen the result in annoying "edge cases". So, after our discussion, I wonder if a minikanren implementation of MIL would overcome the limitations of the Prolog and ASP implementations - even at the cost of expensive occurs-checks etc. I mean, at least if one gives up efficiency, there should be an equally important reward, and an in-practice realisation of the in-principle soundness and completeness of the approach would probably be that. In my opinion anyway. At the very least it's a more compelling reason to re-implement a system in a new language than "well, it's an implementation in a new language". Food for thought, post-thesis :) ____________ [1] Metagol: https://github.com/metagol/metagol (Try to ignore the scary header) [2] Thelma: https://github.com/stassa/thelma [3] Louise: https://github.com/stassa/louise [4] Hexmil: https://arxiv.org/abs/1805.00068
In William Byrd on logic and relational programming, miniKanren (2014)
Statistics, no. Machine learning, yes. The most successful inductive programming approaches (learning programs from data) are Inductive Logic Programming approaches that basically learn Prolog programs from data. Prolog as a language with an automated deductive theorem prover as an interpreter is uniquely suited to learning new prorgams, because, it turns out, induction (learning new theories from observations and existing theories) is the inverse of deduction (explaining observations with existing theories). For some modern examples of learning Prolog programs with Prolog see: 1. Metagol: https://github.com/metagol/metagol 2. Louise: https://github.com/stassa/louise 3. Popper: https://arxiv.org/abs/2005.02259 The first repo has a list of biblio right at the end, though it's a bit too formal maybe. Full disclosure: I started my PhD on Met
Read the rest
agol and Louise is my work. For scientific computing, check out this paper: Using logic programming for theory representation and scientific inference https://www.sciencedirect.com/science/article/pii/S0732118X2... And see repo with code here: https://jeanchristopherohner.github.io/theory-toolbox-2/
In Prolog for Beginners

MAKE YOURSELF AT HOME

A few shortcuts

/
Search the collection
?
Open this guide
Esc
Close dialogs and navigation

Your saved repositories and interests stay in this browser.