Preface: Commonsense knowledge is widely considered to be one of the most difficult issues in AI. In our approach, we introduce a Word-Sensibility Model as a different way to represent commonsense knowledge. As an application, it is a schema and knowledgebase to assist textual analysis. Sensibility is about our relationships to things and the responses that are generated. Consider that there are things we use and don’t use – drawn to or repelled by – make us excited or relaxed. Sensibility is a term often used to denote these kinds of responses. We suggest, that our responses are repurposed where the dynamic sense of them become part of our communicative nature and inform topical as well as contextual orientations. The aim is to present an anatomy for sensibility similar to the way formal logic presents an anatomy for rationality. The word-sensibility model provides an ecological systems perspective on how humans share responsive orientations that factor into commonsense knowing and metaphorical thinking. Theoretically, word-sensibility is not about different ways of reasoning but about different ways of coping. Here, coping reflects the social-dependent nature of the human species, a dependence that language represents. And that speaks to things like, the benefits of learning a foreign language, for instance, or the pressures we feel on our personal attitudes and behaviors as we move from one social group to another. The model implies that what we are, what we know of the world and ourselves, what we believe—isn’t bordered or contained or firm. It’s a cloud and can change easily like a crossfade into another situation. At its core, word-sensibility is that which prescribes internal and external distinctions to contextual units. That is, it aims to model an agents internal responses to external occurrences at different nested levels in the system.

• Introduction …………….…………………..3
• Micro-Topics ……………………………….4
• Quadranyms ……………………………….5-10
• Quadranyms & Polynyms…………….11-12
• Coherent Bias……………………………..13-15
• Ecological Perspective………………….16-19
• Contextual Timelines…………………….20
• Acquisition & Knowledgebase……….21-22
• Commonsense Representation……..23

A Theoretical Look at the Role of Words for AI:

Meta-Lexicography & Commonsense Ontologies

BuildIntuit.com

The Quadranym Model of Word-Sensibility (Q): An Ecological Systems Perspective On Word-Level-Concepts & Contextual Unitizations.

In application, the aim is to assist machine access to lexical data (dictionaries) as pertaining to commonsense situations in text. It is a method concerned with computational lexicography and ontology for AI. It involves meta-lexicography as it aims to deal with accessing word sense by using special paradigms to search words and simulate the human ability to sense dynamic relations between signs.

In Q analysis, the signifier requires two different systems of context to be signified. A distinction is made between the Situational Context (conditions in the world) and the Dynamical Context (umwelt: organismic responsiveness). The model is inspired by an ecological system of agency for accessing knowledge.

Micro-Topics (or word-topics); help organize, characterize or summarize lexical information. They’re pre-textual micro-unit renderings of context used to anchor focus given to words, sentences and to contextual development.

The basic idea is for machines to have the ability to learn to abstract the human experience of contextual expectation, building that abstraction up from the word-level, and in this way machines can make better sense of the things that people typically know. Metaphoric analysis is our primary goal.

In this theoretical approach, we apply a paradigmatic relationship layer to a lexicon in which the data components are related to form micro-topics. The product resembles a thesaurus. The project includes a wiki, schema, source code, various acquisitions & interfaces. Extracting from text is a current goal.

What is a Quadranym?
We define Quadranym as a four-part conceptual construct using a dual-axis Mode-State model. It operates as a virtual unit of orientation and constraint.

Expansion-Reduction (ER-mode) and Objective-Subjective (OS-state).

E = expand: open		O = object: barrier
	N = Topic Name: door
S = subject: passage		R = reduce: close

A Quadranym represents a virtual unit of responsiveness. It is inspired by embodied language processing involving motor and sensory perception and experiential traces.

The Quadranym is a simple organizing construct that is reasonably explained in minutes. Below is an introductory outline of the Quadranym dimensions.

The Prime Dimensions & Meta-Dimensional Role Examples:

• E: Expansive (potential mode) term examples: potential, active, unity, group, over, all, new, implicit, big, learning, playing, expand (or a kind of child like view of life where the world is constantly unfolding), mystery, sustain, spatial openness.
  
• R: Reductive (actual mode) term examples: actual, passive, plurality, it, that, you , me, fact, in, down, measure, explicit, focus, small, reduce (or a kind of adult view of life where the world needs controlling), familiar, deny, spatial closeness.
  
• O: Objective (potential state) term examples: potential, becoming, condition, variant, practice, decide between, interpersonal, social, object, temporal endings.
  
• S: Subjective (actual state) term examples: actual, being, coherent, core, constant, perspective, beliefs, desires, intrapersonal, subject, temporal beginnings.

Various metaphysical notions apply to the Quadranym prime dimensions, such as, part/whole, plurality/unity, being/becoming, time and space. Note that actual-state refers to a unit’s actual state of being. This mainly refers to the context of being, core sense and self-centered bias. When referring to physical properties and laws, actual is a situational context. We call this The Objective Field, i.e., real-world entities. A dynamical contextual orientation virtually seeks to optimize its relations to the objective field. Objective and subjective states refer to ontological categories of context and not any division real in nature.

Active sense initiates a micro-topics function and argument. Passive sense is when the function and argument is finally configured for the situation. Actual state is the source. Potential state is the target.

Hemisphere relations provide the general template configuration for representation:

• Left Side, Superset-Active, E_Mode(s_state) = E_Active_Potential(S_active_actual)
• Right Side, Subset-Passive, R_Mode(o_state) = R_Passive_Actual(O_passive_potential).

space(x) → Infinite{…)(void{…}) ⊇ Finite{…}(between{…})x

door(x) → Open{…)(passage{…}) ⊇ Close{…}(barrier{…})x

• The superset is the source condition.
• The subset is the target variable.

Note: the Source Condition is also called the Coherent Bias.

Independent & Dependent Variables.

In this particular Quadranym, space anchors on infinite dependent on finite, on between as the target variable (a.k.a. condition potential) and on void as the zero-point.

Dependent Potential Infinite	Independent Actual Finite	Condition Potential Between
occupants	capacity	room
growth	time	size
far	near	distance
open	close	door

Above, occupants is dependent on capacity to target room. Growth is dependent on time to target size. The next two examples seem to be interchangeable. However, because Quadranyms are nested units this is correct. Far and open are nested under infinite. Near and close are nested under finite. The source condition of the spatial unit is referenced on void. If there is no other condition then void is the infinite condition otherwise between is the target condition.

Consider climate inferring door utility. Text: ”Close the door because it’s cold outside.”

Any variability in climate is dependent on the division of spatial modes. All space = all temperature non-discrete. In the door realm, any variance is dependent on barrier. All Quadranyms can conform in this way in some manner.

To vary climate, open is dependent on closed.

barrier = climate variance

Because door is a subset of the spatial domain it is nested in spatial dimensions as illustrated in the matrix of nested spatial topics below. The matrix represents a Polynym. Each topic name represents one of its dimensions (ranks). Polynyms have any number of dimensions.

• Conceptnet relations, e.g., between isA barrier, open motivatedBygoal out.
• Also, relations can cross rank topic terms, e.g., out obstructedBy barrier.

Although the matrix provides word relations, it’s less about word relations and more about the dynamic factors for word relations.

Consider the assertion, “Science makes predictions.”

Several topics come together to form a Polynym. The terms in each of the (EROS) columns tend to fill in a tantamounting or pervading sense.

A Quadranym knowledge base aims to add a supplemental layer to any knowledge base that provides word relations (directed edges). Specifications come from WordNet or other lexicons.

The coherent bias generally represents a heuristic bias or experiential trace. Essentially, it is an instance of coherent sense. In the model, there is a distinction to be made between the coherent sense of a word and the potential conditions it may aim at in the world.

Intimating Mental States.

(∀x) eat(x) ⟹ [Sate(hungry) ⊇ Starve(food)x]

• Modes provide measure
• States provide arguments
• hungry anchors orientation
• food is the target variable

In the model, once motivated, the listener’s intention is to find cues in the content so to sync with oscillating coherent and conditional factors.

What is required here is the most basic topic and its source condition, a condition that the agent and the patient can share. Any other knowing requirements involves nesting more micro-topics.

Micro-topics are virtual orientations shared between people. Just like there are different word senses, there are different orientations of micro-topics. Consider the different orientations for the topic eat below.

Topic Name: Eat(x)	Weights
(∀x) eat(x) ⟹ [Sate(hungry) ⊇ Starve(food)x]	1.0
(∀x) eat(x) ⟹ [Intact(chew) ⊇ Fragment(substance)x]	0.6
(∀x) eat(x) ⟹ [Available(consume) ⊇ Deplete(resource)x]	0.5
(∀x) eat(x) ⟹ [Stable(corrode) ⊇ Disintegrate(material)x]	0.1

Check the system page for information on the coherent bias and data clustering.

In the model, a conservation of order is virtually provided by micro-topics. In other words, a conflict between coherent and conditional factors will virtually generate new micro-topics. When people have conflicting responses to new input, there is a social motivation to produce like minded objectives. This is a reciprocal relationship between people to influence outcomes in the world. The aim is to represent these kind of relations between intra-subjective units.

Virtual Perspective: Each Quadranym represents a discrete dynamical system. All systems draw upon the environment and then (in response) give back to it, thus participating in an ecology of dynamical systems.

Units when linked together form scripts. The system generally consists of Units, Scripts and Layers.

Scripts can run linearly or can be broken into contextual timelines (layers) and run simultaneously. These structures are saved and repurposed in new modal systems.

Upper scripts constrain lower scripts.

Unattended & Attended Layers  → Goal:

General Layer:[Urge(survive) → Resolve(nutrition)]

General Layer:[Urge(nutrition) → Resolve(hunger)]

General Layer:[Urge(hunger) → Resolve(food)]

Specific Layer:[Urge(food) → Resolve(water)]

Specific Layer:[Urge(water) → Resolve(fish)]

 

In this approach, what implicates a sensibility are the skills and resources available to address new developments. Environments change behaviors and these changes are represented in micro-units. A reciprocal causation between an organism and its environment:

• Representing environmental resources is a flux.
• Representing organismic resources is a unit.

1.flux is double brackets: [b] → [a]

2.unit is single brackets: [a → b]

• flux: a dynamic sense driven by the environment.
• unit: how that sense has been driven before.

Ultimately, the process depends on the strategic ability to layer scripts into Contextual Timelines.

• Contextual Timelines refer to procedural constraints e.g., when you get up for work; take shower, brush teeth, have breakfast, rush out etc… work forms the contextual timelines.

Hierarchical layers represent Contextual Timelines. Some layers cycle more generally and others more specifically. The more new units appear in a layer the more specific it is to a task. Strategies may supplement or replace layers or rearrange hierarchical orders. Procedural layers are mixed and matched to optimally describe an agents condition of responsiveness given the occurrent situation. The primary advantage of layers is the ability to create various hierarchical orders. In this way, an act can be constrained by different motivations.

Consider for instance a chimpanzee who smashes small stones with a big rock. Maybe the act continues just for the dynamic sense of it, the impact and stones fractured into fragments. Now consider the act repurposed to crack nuts. A new motivation, a new system of responsiveness is organized.

We introduce a method to acquire commonsense and interdisciplinary knowledge. It is a practice of collecting and organizing idea sets: sets of terms, their area, their source and their relations. Essentially, an idea set is a set of terms representing a certain number of dimensions used to strategically unfold, frame or simplify a concept. A dimension number is called a polynym i.e., mononym, duonym, trionym, tetranym, pentanym and so on. A collection of idea sets resembles a kind of thesaurus. The content can be used to help populate knowledgebase systems and enhance queries.

Quadranyms and polynyms when pertaining to system processes form a matrix that represents a general and overarching cognitive framework for a ecological systems approach for an ontology of commonsense representations providing dynamic factors.

Below is a basic Quadranym Interface. There are a variety interfaces and games developed and more being developed for the project. Phrasal templates are created and extracted to help fashion and test Quadranym effectiveness. Special realms are used for polynym matrices and scripts. Existing polynyms are being collected. A current goal is to develop extraction methods using Quadranym realms to cluster terms.

As a public interface, Quadranyms are represented on y mode and x state axes.

To conclude, the aim was to bring some understanding to word-sensibility theories and components. The full idea involves unpacking a multi-organizational dynamic system. The model is a method for commonsense representation that introduces the idea of motivated-dynamical-contexts anchoring word-level concepts, we refer to it as Word-Sensibility. Any word in the system can be considered a motivated-dynamical-context. Central to this process is the idea that there is the immanent and what necessarily must transcend the immanent, by which this implicates the skills, volition and resources that one has to cope in the world. Word-sensibility is proposed as a hypothetical construct pertaining to affective and conative components. It’s about social instincts, habits, emotions and volition reinforced by the environment. The primary advantage of standardizing a model of word-sensibility is that experiential myths, or metaphorical mappings about experience may be characterized using universal themes in a domain general framework. The current goal is to build a knowledgebase populated with basic thematic realms such as, spatial, temporal, metabolical, emotional and social.

Check out About Page and System Page for more information.

Contact us. We look forward to any questions and suggestions.

Creative ambitious developers welcomed, contact us!

Thank You :)