RE: Emergent Semantics

Semantic Web Interest Group,
Artificial Intelligence Knowledge Representation Community Group,
Paola Di Maio,

The specific scenarios that the data structures were devised for include modeling natural language and narrative. The data structures can be made more general-purpose. Utilizing parametric polymorphism – generics – the indicated graph-based data structure can generalize from scenarios involving the interrelating of events to scenarios involving the interrelating of arbitrary other data structures.

The preliminary interface could generalize to:

public interface IGraph<TNode, TEdge> : IIndividual
{
    //string Id { get; set; }
    //IGraph<TNode, TEdge> GetSubgraphById(string id);

    IGraph<TNode, TEdge> Root { get; }

    IQueryable<TNode> Nodes { get; }
    IQueryable<TEdge> Edges { get; }
    IQueryable<IGraph<TNode, TEdge>> Subgraphs { get; }

    TNode FindOrCreateNode(/*...*/);
    TEdge FindOrCreateEdge(TNode source, PredicateExpression relation, TNode target);
    IGraph<TNode, TEdge> FindOrCreateSubgraph(IEnumerable<TEdge> edges);

    bool Add(TEdge edge);
    bool Remove(TEdge edge);
    bool Replace(TEdge remove, TEdge add);
    bool Contains(TEdge edge);

    IList<CompoundExpression> Semantics { get; }
}

We could also utilize two interfaces, as per:

public interface ISubgraph<TNode, TEdge> : IIndividual
{
    string Id { get; set; }

    IGraph<TNode, TEdge> Root { get; }

    IQueryable<TNode> Nodes { get; }
    IQueryable<TEdge> Edges { get; }

    bool Contains(TEdge edge);

    IList<CompoundExpression> Semantics { get; }
}

public interface IGraph<TNode, TEdge> : IIndividual
{
    ISubgraph<TNode, TEdge> GetSubgraphById(string id);

    IQueryable<TNode> Nodes { get; }
    IQueryable<TEdge> Edges { get; }
    IQueryable<ISubgraph<TNode, TEdge>> Subgraphs { get; }

    TNode FindOrCreateNode(/*...*/);
    TEdge FindOrCreateEdge(TNode source, PredicateExpression relation, TNode target);
    ISubgraph<TNode, TEdge> FindOrCreateSubgraph(IEnumerable<TEdge> edges);

    bool Add(TEdge edge);
    bool Remove(TEdge edge);
    bool Replace(TEdge remove, TEdge add);

    bool Add(IEnumerable<TEdge> edges);
    bool Remove(IEnumerable<TEdge> edges);
    bool Replace(IEnumerable<TEdge> remove, IEnumerable<TEdge> add);

    bool Contains(TEdge edge);

    IList<CompoundExpression> Semantics { get; }
}

In the indicated source code, a CompoundExpression is a predicate calculus expression – the semantic data upon events, graphs and subgraphs are sets of predicate calculus expressions.

Design topics include providing access to arbitrary subgraphs, for instance by specifying sets of edges. The method FindOrCreateSubgraph,

    ISubgraph<TNode, TEdge> FindOrCreateSubgraph(IEnumerable<TEdge> edges);

looks up a subgraph, based upon its composition as a set of edges, and, if the data structure cannot find an existing subgraph data structure, it creates a new one to return. The graphs and subgraphs can be full-fledged objects, with properties, and, with:

    IList<CompoundExpression> Semantics { get; }

one can attach semantic data to them. That is the gist: to be able to access arbitrary subgraphs of a graph, either to get or set properties upon them or to access semantic data upon them.


Best regards,
Adam

________________________________
From: Paola Di Maio <paola.dimaio@gmail.com>
Sent: Sunday, May 26, 2019 12:24:58 AM
To: Adam Sobieski
Cc: ProjectParadigm-ICT-Program; semantic-web@w3.org; public-aikr@w3.org
Subject: Re: Emergent Semantics

Dear Adam
thanks for sharing

please give examples of what to do with this, and what benefit it brings

thanks
PDM


On Sun, May 26, 2019 at 8:27 AM Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>> wrote:
Semantic Web Interest Group,
Artificial Intelligence Knowledge Representation Community Group,
Milton Ponson,

I have modeled, in a preliminary manner, the recursive tree-based and graph-based data structures which I previously described in this thread.

In the file: https://github.com/AdamSobieski/Logic/blob/master/Logic/Collections/ISemanticTree.cs , one can observe a tree node data structure such that semantic data can be attached. In the context of natural language, such data structures could be of use for modeling lexemes, phrases, sentences, paragraphs and so forth. The attached semantic data can be described as, at least in part, emergent – sentences, for instance, can have more semantics than the sum of the semantic data of their parts. The contents of the semantic data upon such tree nodes could be populated by processes including varieties of inferencing which occur during reading comprehension.

In the file: https://github.com/AdamSobieski/Logic/blob/master/Logic/Narrative.cs , one can observe event nodes, event edges and event graphs. Events can have semantic data attached to them. Event graphs, graphs of interrelated events, can have semantic data attached to them. Event graphs are modeled in such a way that arbitrary event subgraphs can be found or created by specifying sets of edges. Events, event graphs and arbitrary event subgraphs can each have semantic data attached to them. The contents of the semantic data upon such events, event graphs and arbitrary subgraphs could be populated by processes including varieties of inferencing which occur during reading comprehension.


Best regards,
Adam Sobieski

________________________________
From: ProjectParadigm-ICT-Program <metadataportals@yahoo.com<mailto:metadataportals@yahoo.com>>
Sent: Tuesday, May 21, 2019 9:56:07 AM
To: Adam Sobieski; semantic-web@w3.org<mailto:semantic-web@w3.org>; public-aikr@w3.org<mailto:public-aikr@w3.org>
Subject: Re: Emergent Semantics

IMHO the following is a viable starting point for a broad discussion on semantics and natural language.

Statement:"language deals with domains of discourse that allow inner reflection, or transmitting or recording about domains of discourse"

From this starting point we can unify Buddhist logical systems, philosophy, religion, science, logic and mathematics in the process showing that they are all subject to Godel's Incompleteness Theorem.

The only way out of this is to strive for consistency using generalized conceptual frameworks such as can be created by using category theory.

Buddhists, linguists, neuro-scientists and psychologists are  slowly converging in their efforts to arrive at cognitive architectures and formal models for consciousness.

And what is emerging is the means to structure AI and AGI in a truly universal way, beyond mere machine learning.

Milton Ponson
GSM: +297 747 8280
PO Box 1154, Oranjestad
Aruba, Dutch Caribbean
Project Paradigm: Bringing the ICT tools for sustainable development to all stakeholders worldwide through collaborative research on applied mathematics, advanced modeling, software and standards development


On Saturday, May 18, 2019 9:37 PM, Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>> wrote:


Semantic Web Interest Group,
Artificial Intelligence Knowledge Representation Community Group,

Greetings. I would like to broach, for discussion, emergence and emergent semantics.

We can envision data structures, e.g. trees and graphs, which contain recursive substructures, e.g. subtrees and subgraphs, such that each substructure can be adorned with semantic content. We can refer to this content which adorns structures and substructures as emergent semantics.

An example scenario is that of natural language where sentences are comprised of lexemes and where sentences are thought to have more semantic content than the sum of the semantic content of the lexemes. Another example is that of narratives, comprised of events, where narratives are thought to have more semantic content than the sum of the semantic content of the events which comprise a fabula.

In the example of natural language, we can envision lexeme data structures as being adorned with semantics (collections of triples, quads or predicate calculus expressions) which indicate the meanings of the lexemes. We can envision phrases, containing and comprised of the lexemes, as being adorned with semantic content. We can envision sentences, containing phrases and lexemes, as being similarly adorned. Paragraphs, comprised of sentences, may, too, be adorned with emergent semantics—and so on.

In the example of narratives, we can envision event data structures as being adorned with semantics which indicate the events’ meanings/contents. We can envision graphs of interrelated events as being adorned with emergent semantic content. We can envision graphs containing subgraphs of interrelated events – each structure and substructure adorned with emergent semantics.

Is there any interest here in emergent semantics? Could anybody recommend any hyperlinks or publications?


Best regards,
Adam Sobieski