The Search Space

The Search Space

By Felix HolmgrenSociety & CultureTechnologyPhilosophy
Download on the App Store

The Search Space episodes

  • #6 - Adam Smith on Answer Set Programming

    Adam Smith talks about Answer Set Programming, and how he's used it in game design and other areas.

    Visit the show's web page: thesearch.space

    Show notes

    | Torsten Schaub at Potsdam University

    Several of his presentations contain the formula

    ASP = DB + LP + KR + SMT^n

    ASP: Answer Set Programming
    DB: Database
    LP: Logic Programming
    KR: Knowledge Representation
    SMT: SAT Modulo Theories

    SAT solver: Boolean Satisfiability

    | this refraction game that I had made a puzzle generator for

    Described in Adam's dissertation.

    | the Potassco tools from University of Potsdam.

    potassco.org

    | A book called Procedural Content Generation in Games

    www.pcgbook.com

    | There's this idea due to Kate Compton ... of the 10 000 bowls of oatmeal problem

    http://www.galaxykate.com/blog/generator.html

    | We modeled a design space of information visualizations, line plots and scatter plots and bar charts

    "Formalizing Visualization Design Knowledge as Constraints: Actionable and Extensible Models in Draco"

    | a Python library called Clorm .. so that you can define your problem instances with Python objects

    github.com/potassco/clorm

    | the concept of elaboration tolerance that when someone gives you some clarification about how your domain works, can you incorporate that change by just adding more code to add more choices or more constraints?

    John McCarthy's paper on elaboration tolerance

    | a paper from Google DeepMind ... describing a system called the Apperception Engine

    "Making sense of sensory input", Evans et al

    | taught a class called Applied ASP a few years ago

    Lecture slides, programming assignments, reading assignments here: canvas.ucsc.edu/courses/1338

    | Once you want to write a program that's more than 30 lines long, how do you profile it? How do you test it? How do you deploy it?

    Slides from a talk about Adam's test automation framework: Unit Test Automation for ASP with Ansunit

    Try ASP from your browser: https://potassco.org/clingo/run/

    ★ Support this podcast ★
    1 hr 14 min
  • #5 - Kevin Feeney on TerminusDb and the Semantic Web

    Visit the show's web page: thesearch.space

    Show notes

    I first became aware of Kevin through a series of blog posts that explain the similarities and differences between these different kinds of databases

    Graph Fundamentals — Part 1: RDF

    Graph Fundamentals — Part 2: Labelled Property Graphs

    Graph Fundamentals — Part 3: Graph Schema Languages

    Graph Fundamentals Part 4: Linked Data

    Then I found out about TerminusDb

    https://terminusdb.com/

    "a bunch of Swedish hackers with a bunch of JSON blobs"

    https://neo4j.com/

    Full quote:

    [...] there have been many more incoherent standards and initiatives that have come out of the W3C’s standards bodies — almost all of which have  launched like lead balloons into a world that cares not a jot.  Nevertheless, it is important to recognise that, hidden in all the  nonsense, there are some exceptionally good ideas — triples, URL  identifiers and OWL itself are all tremendously good ideas in essence  and nothing else out there comes close. It is a sad testament to the  suffocating nature of design by standards committee which has consumed  countless hours of many thousands of smart and genuine researchers, that ultimately the entire community ended up getting it’s ass kicked by a  bunch of Swedish hackers with a bunch of json blobs — the Neo4j property graph guys have had a greater impact upon the real world than the whole academic edifice of semantic web research.


    The Semantic Web as a movement came out of Tim Berners-Lee

    One of the seminal articles:

    https://web.archive.org/web/20171010210556/https://pdfs.semanticscholar.org/566c/1c6bd366b4c9e07fc37eb372771690d5ba31.pdf

    The Semantic Web is not a separate Web but an extension of the current one, in which information is given well-defined meaning, better enabling computers and people to work in cooperation. ... Adding logic to the Web—the means to use rules to make inferences, choose courses of action and answer questions—is the task before the Semantic Web community at the moment.

          — May 17, 2001, The Semantic Web - A new form of Web content that is meaningful to computers will unleash a revolution of new possibilities, Tim Berners-Lee, James Hendler and Ora Lassila 

    The standardization of RDF, the standardization of OWL


    the big, big gap in a lot of the standards of the semantic web was some type of closed world reasoning regime


    I was one of the developers of software that ran a thing called Indymedia back in the early two-thousands.


    most of the impetus for OWL came out of the description logic community

    ...there's a number of very well-known and very accomplished description logic people [...] like Peter Patel-Schneider and [Ian] Horrocks in Oxford.


    ...they were using predicates to point out that two things, two data structures are the same thing. But the standard didn't ... mean for that to be used for things that just happened to be the same real world thing...

    Kevin discusses this misuse of owl:sameAs and owl:equivalentClass in the the fourth of his blog posts linked to above.

    I was talking to some of the guys in Semantic Arts, who are very busy and active consultants in the area.


    And then there is this thing called RDF stores...


    [RDF stores] are based around this concept of a triple - predicate subject object


    Just like Google do actually on their front page now for their knowledge graph

    The Google Knowledge Graph was introduced in 2012 with the great slogan "things, not strings"

    ...the other thing that triples have ... is it makes revision controlled databases possible. 

    From TerminusDB: A Technical History:

    ...we adopted a delta encoding approach to updates as is used in source control systems such as git. This provides transaction processing and  updates using immutable database data structures, recovering standard  database management features while also providing the whole suite of  revision control features: branch, merge, squash, rollback, blame, and  time-travel...


    I've actually seen that very thing being described as a benefit of property graphs that each relation has its own ID and it can have [a] whole data structure associated with it

    See for example neo4j's blog post RDF Triple Stores vs. Labeled Property Graphs: What’s the Difference?, in the section "Difference #1: RDF Does Not Uniquely Identify Instances of Relationships of the Same Type".

    You can do it in SQL these days, but it's sort of a later addition... the WITH syntax, Common Table Expressions they are called... you can actually do recursive queries. 

    I once showed up at a neo4j meetup with some examples of doing graphy queries in PostgreSQL using Common Table Expressions. (The presentation would have been more impressive if I had but some indexes on those tables...) A way better introduction is the excellent page on The WITH Clause in the SQLite documentation.

    People beat up on normal-form modeling and SQL way more than they should.

    A good recent blog post on this topic: Normalization is not a process.

    Even when I'm modelling graph stuff, I start off with basically
    1 hr 28 min
  • #4 - Markus Triska on constraints and the power of Prolog

    Visit the show's web page: thesearch.space

    Show notes

    The Power of Prolog, Markus' ongoing book project

    "It says, 'Find your way from darkness to light,' which is one of the quotes that occurs in The Knight of Cups, which is a recent movie by Terrence Malick."


    "I was introduced to Prolog by Ulrich Neumerkel at the Vienna University of Technology."


     "the convener of the Prolog ISO standard group."


    "He has developed his own teaching environment called GUPU, which means talk-assisted programming environment [Gesprächsunterstützende Programmierübungsumgebung]."


    Declarative program development in Prolog with GUPU (paper)

    "one fellow student came to me and asked for a solution to [what] you may now know as the SEND MORE MONEY task."

    SEND + MORE = MONEY

    "People have ... in different languages, explored multi-dispatch, multi-methods, and all kinds of stuff."


    "... the notion that an algorithm can be decomposed into a logic aspect and a control aspect. So this is written as 'algorithm is logic plus control.'"


    "...let's look at the HTTP framework of SWI-Prolog"


    "...the first use case that Prolog should solve was translating weather reports between French and English. It was the METEO system..."


     "For this purpose ... DCGs were invented. So this is a sub-formalism in Prolog that lets us describe lists in a very natural way."


    "...you have kind of a special syntax with the keyword is, and many of the normal, nice properties of Prolog start breaking down."

    (is)/2  is not actually a keyword or a special syntax. Markus describes the issues involved very well here:

    https://github.com/triska/clpz#an-impure-alternative-low-level-integer-arithmetic

    and here:

    https://www.metalevel.at/prolog/clpz

     "It is abbreviated as CSP: Constraint Satisfaction Problem."


    "You have variables, you have associated domains, which in plain Prolog are always Herbrand terms, named after Jaques Herbrand."


    "In the literature, this has even been mentioned as the Holy Grail of computer science. This property [or] at least the idea that users or application programmers specify a task and the system solves it."


    Eugene  C  Freuder.  “In  pursuit  of  the  holy  grail”.  In:Constraints 2.1 (1997), pp. 57–61. (not freely available)

    There is an ongoing series of workshops following up on the ideas in the original paper:

    https://freuder.wordpress.com/pthg-20-the-fourth-workshop-on-progress-towards-the-holy-grail/

    "For comparison, [...] the constraint solver for SICStus Prolog is hundreds of times faster than for example what I've implemented."


    > "rusty-wam, implemented in Rust"
    (hence renamed Scryer Prolog, with Markus being an active contributor)

    > "There is O-Prolog, which is implemented by Kenichi Sasagawa from Japan."

    > "..in the case of Tau Prolog, it's implemented in JavaScript and they can embed Prolog in a web browser"


    The Power of Prolog YouTube channel

    ★ Support this podcast ★
    1 hr 6 min
  • #3 - Chris Martens on narrative generation

    Visit the show's web page: thesearch.space

    Show notes

    Chris Martens' academic website
    https://www.csc.ncsu.edu/people/crmarten

    04:30
    "Programming Interactive Worlds with Linear Logic", Chris' Ph.D. thesis

    06:10
    James Meehan’s, Tale-Spin thesis
    "The Metanovel: Writing Stories by Computer"

    A great post about the story of Tale-Spin's creation:
    https://grandtextauto.soe.ucsc.edu/2006/09/13/the-story-of-meehans-tale-spin/

    18:40
    The Twelf Project

    "a language used to specify, implement, and prove properties of deductive systems such as programming languages and logics"

    The dependently typed logif LF


    20:50
    Linear logic

    Jean-Yves Girard

    The original paper
    (The first sentence begins: "Linear logic is a logic behind logic...")

    22:00
    "A form of logical implication pronounced A lolly B" ... I wish I had a screen to draw on"
    Looks like this:  A -o B  (a modified arrow from A to B)

    25:10
    The frame problem

    25:20
    Temporal logic
    Event calculus

    26:50
    Pandemic board game

    "5% of my design royalty for Pandemic products is donated directly to Doctors Without Borders"

    https://www.leacock.com/about

    44:40
    Interactive fiction

    "software simulating environments in which players use text commands to control characters and influence the environment"

    https://www.ifarchive.org/

    47:30
    Behavior trees

    56:00
    R. Michael Young

    "His research focuses on the development of computational models of interactive narrative with applications to computer games, educational and training systems and virtual environments."


    57:30
    https://twitter.com/chrisamaphone
    github.com/chrisamaphone

    ★ Support this podcast ★
    1 hr 1 min
  • #2 - Ryan Brush on retaking rules for developers

    Show notes (full notes coming up, please check back soon)

    Ryan's talk about Clara at Strange Loop (2014)
    https://www.youtube.com/watch?v=Z6oVuYmRgkk


    https://github.com/quoll/naga
    https://github.com/ulfurinn/wongi-engine
    https://github.com/jruizgit/rules

    ★ Support this podcast ★
    50 min
  • #1 - The Poet of Logic Programming

    Show notes

    08:10
    John Alan Robinson
    https://en.wikipedia.org/wiki/John_Alan_Robinson

    Computational Logic: Memories of the Past and Challenges for the Future
    http://www.computational-logic.org/iccl/downloads/Robinson-CL2000.pdf

    Maarten van Emden has great material about Robinson, including two interviews:

    • https://vanemden.wordpress.com/2016/09/16/alan-robinson/
    • https://vanemden.wordpress.com/2010/06/08/interview-with-alan-robinson-inventor-of-resolution-logic/
    • http://aarinc.org/Newsletters/089-2010-10.html#robinson


    08:55
    The Early Years of Logic Programming (Kowalski)
    https://www.doc.ic.ac.uk/~rak/papers/the%20early%20years.pdf

    12:05
    The British Nationality Act as a Logic Program
    https://www.doc.ic.ac.uk/~rak/papers/British%20Nationality%20Act.pdf

    12:45
    Horn clause logic
    https://en.wikipedia.org/wiki/Horn_clause

    14:32
    Dual process theory
    https://en.wikipedia.org/wiki/Dual_process_theory
    https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow

    16:30
    Alternatives to logic representations of knowledge

    • https://en.wikipedia.org/wiki/Semantic_network
    • https://en.wikipedia.org/wiki/And%E2%80%93or_tree
    • https://www.doc.ic.ac.uk/~rak/papers/History.pdf


    17:10
    Minsky’s frames
    https://en.wikipedia.org/wiki/Frame_(artificial_intelligence)

    17:30
    Non-monotonic logic
    https://en.wikipedia.org/wiki/Non-monotonic_logic

    18:10
    First-order logic
    https://en.wikipedia.org/wiki/First-order_logic

    20:45
    List concatenation (the append/3 predicate)
    http://www.learnprolognow.org/lpnpage.php?pagetype=html&pageid=lpn-htmlse24

    23:55
    Datalog
    https://en.wikipedia.org/wiki/Datalog

    Answer Set Programming
    https://en.wikipedia.org/wiki/Answer_set_programming

    25:50
    Cordell Green
    https://en.wikipedia.org/wiki/Cordell_Green

    26:50
    "it was Cordell who was interested in knowledge representation"

    STANFORD ARTIFICIAL INTELLIGENCE PROJECT, MEMO Al-96
    The Application of Theorem Proving to Question-Answering Systems
    https://www.kestrel.edu/home/people/green/publications/green-thesis.pdf

    Theorem-Proving by Resolution as a Basis for Question-Answering Systems
    https://www.kestrel.edu/home/people/green/publications/theorem-proving.pdf

    29:50
    Planner, Hewitt, Winograd
    https://en.wikipedia.org/wiki/Planner_(programming_language)
    https://en.wikipedia.org/wiki/Terry_Winograd

    35:00
    Monads
    https://en.wikipedia.org/wiki/Monad_(functional_programming)

    40:10
    Alain Colmerauer
    https://en.wikipedia.org/wiki/Alain_Colmerauer

    Un Systeme de Communication Homme-Machine en Francais
    http://alain.colmerauer.free.fr/alcol/ArchivesPublications/HommeMachineFr/HoMa.pdf

    42:00
    Pat Hayes
    https://en.wikipedia.org/wiki/Pat_Hayes
    https://www.ihmc.us/groups/phayes/

    43:15
    The Birth of Prolog (Alain Colmerauer and Philippe Roussel)
    http://alain.colmerauer.free.fr/alcol/ArchivesPublications/PrologHistory/19november92.pdf

    46:00
    Computational Logic and Human Thinking: How to be Artificially Intelligent
    https://www.doc.ic.ac.uk/~rak/papers/newbook.pdf

    49:00
    The Pyramid Principle: Logic in Writing and Thinking (Barbara Minto)
    https://www.goodreads.com/book/show/1990595.The_Pyramid_Principle

    51:00
    Unification
    https://en.wikipedia.org/wiki/Unification_(computer_science)

    52:20
    Backtracking
    https://en.wikipedia.org/wiki/Backtracking

    56:10
    Production rules
    https://en.wikipedia.org/wiki/Production_system_(computer_science)

    57:10
    Expert systems
    https://en.wikipedia.org/wiki/Expert_systems

    01:08:30
    The Fifth Generation Computer Systems project
    https://en.wikipedia.org/wiki/Fifth_generation_computer

    1:09:50
    MapReduce
    https://en.wikipedia.org/wiki/MapReduce

    1:17:00
    Event calculus
    https://en.wikipedia.org/wiki/Event_calculus

    1:18:20
    Logic Production Systems
    http://lps.doc.ic.ac.uk/

    1:19:40
    Logical Contracts (company)
    http://logicalcontracts.com/

    1:29:45
    Minsky's attack on the perceptron
    https://en.wikipedia.org/wiki/Perceptron

    ★ Support this podcast ★
    1 hr 36 min
  • Trailer

    Logic Programming, you say? That might sound like either an oxymoron or a tautology to you, depending on who you are. If you have heard about Logic Programming before, you are probably thinking about Prolog - a programming language that is almost 50 years old, and which many find fascinating but also frustratingly limited.

    What is logic? Many people consider it synonymous with dry intellect, humorless bureaucracy, the opposite of creativity. But in my understanding, logic is an attempt to capture the essence of thought, which is to say the essence of what we humans find reasonable, persuasive, and possible. The study of logic goes straight to some of our fundamental intuitions about the world and ourselves, and it's really not clear at all what they are rooted in. People have lots of different opinions about this. Logic looks very clear-cut on the surface, but when you dig you bump into some of the deepest perennial questions in philosophy. And this doesn't change when we arrive at modern, formalized logic, which sort of looks like mathematics - you are still dealing with deep questions about human psychology and cognition, and how we perceive the physical world.

    Now, it turns out formal logic can be used as a programming language. If you are careful about which parts of logic you use, and in which form you write things down, formal logic fits very well with how programming languages are executed. So you can write down some logical formulas, and you get two interpretations: the logical interpretation, and the procedural interpretation which is how you run it on a computer. And these two are, let's say, in sync - you get the deductions you would expect from both interpretations.

    As a programming paradigm, this has some really unusual and fun features, and since logic in the first place is, again, an attempt to capture the essence of human though, Logic Programming is in many respects very intuitive. To me, it's the closest to what I imagined programming would be like when I was 10 and had just started to understand that there was a way to tell computers what to do.

    ★ Support this podcast ★
    6 min

About The Search Space

From the publisher's feed

The Logic Programming podcast. About the history, future, and wider landscape of Logic Programming.