Structure vs Semantics
The Semantic Perspective Project is an entirely new approach to information and knowledge representation, where information composition is done via “meaningful perspectives“: what the meaning of a concept is, when viewed from the perspective of a different concept, hence the project’s name.
Knowledge representation is essential to any computational system that operates with information as opposed to data, from traditional expert systems to modern concepts like the semantic web. The distinction between the two terms “data” and “information” is critical, because in spite of the fact that we have created systems that are highly efficient in storing and manipulating huge amounts of data, we are yet to create a single system (represent, store, manipulate) that can even marginally tackle the kind of information a person does on a daily basis.
The Semantic Perspective Project arose from the constructive quest to identify why current knowledge representation and manipulation systems are incapable of fulfilling the “semantic promise” and to provide, or at least try to provide, an alternative solution that works around the identified deficiencies.
Comparison with mainstream ontologies
Although there is much more to the topic that can be presented in a one pager, I will try to do a side by side comparison of some of the core aspects the two models. On the left we have the mainstream approach, the approach employed by all ontology models regardless of how different they are, and on the right column we have the semantic perspective approach.
| Mainstream Approach | Semantic Perspective Approach |
| Structure defines the concept. In the mainstream approach the accent is on structure. Concepts are defined by a set of characteristics and these sets of characteristics are taken as being universal to all elements of a certain concept. For example a Person is a {first name, last name, date of birth, etc.} n-tuple, no more, no less. | Concept is independent from structure. Within the semantic perspective the structure is irrelevant. A concept is a structure less reference to some counterpart in reality. Concepts may have specifics (materialization of concepts) but these are but references and they are free to participate in any construct, thus are able to have an infinity of variations of structures or none at all. |
| Reality is classified in Types. The entire semantic reality is a set of elements and their relations, constructed on the basis of a blueprint which is a predefined collection of named structures: Person, Book, etc. etc. which predates reality. As such, the actual physical reality that we try to represent must fit into this predefined set of structures, whether this is possible or not. | Classifications are part of the reality, types emerge but they do not control reality. Reality is free to have any form, no assumptions of its form is made a-priori. Characteristics are represented by sets of patterns that emerge for certain concepts, but they do not lock in a concept. Concepts and specifics are free to participate in any number of sets of patterns (aka Types). Types hint towards the possibility of a specific to have certain characteristics, but this is not a mandatory fact. |
| Relations are part of the definition. As part of reality’s definition, the mainstream approach employs relations. Relations are predefined concepts that involve the two or more types. This implies that instances of types can exist in such and such relation, for example “member_of”, “friends_with”, etc … For two instances to exist in any kind of relation, such relation must be part of the definition. | Relations do not exist in the standard sense. The semantic perspective contains only two meta-relations: the divergence (to create perspectives) and the correlation (to connect perspectives).There are no predefined relations the way “friends_with” would be in the classic sense. Concepts and specifics can be part of any number of construct which means there is no limit on the number of relations a concept or specific may be part of. Relations are part of the reality not the definition of reality. |
| Nomenclature is open, terms can be invented. Terminology is part of the definition, and the creator of the definition is free to use any name and associate it with any concept. Ex: The Person can be named Person or Nosrep (person written backwards), there is simply no limit on how things should be named.This terminology invention is especially true in the case of relations (“member_of”, “friends_with”, etc …), where such terms are associated with strict definitions that have a predefined meaning within the definition suite. In some instances, thousands of new words appear in ontologies which may or may not be the same as the dictionary meaning of the word. At the same time, the same word appears on different ontologies, a deceptive thing, because the two names are not in fact compatible. | Nomenclature is restricted to the dictionary. The semantic perspective presumes that humans have already created a name for all concepts in their reality and gathered them up in a dictionary for common reference. The semantic perspective approach takes this millenia long effort very seriously and abides to it. If something is termed “Person” then this has got to mean exactly what the dictionary says.Terminology invention is discouraged because it is a great barrier in the adoption of the definition. Humans already have a complete set of terminology. Why should there be other parallel ones with partial overlapping. |
| Ontologies define the reality. The totality of types, relations, rules that form an ontology represent the seed of a very specific reality that has its roots in the ontology. Every instance is a materialization of a type or of a relation that is predefined in the ontology and the totality of these instances form this emerged reality. Similarly to how an instance is a manifestation of a type in the world of these ontologies, realities are materializations of ontologies. These realities however are almost always incompatible amongst eachother, because they are based on different definition sets. | Ontologies define particular views reality. In the semantic perspective model, the ontologies do not exist in the same sense as they do in todays semantic computing world. The SPInDL ontology is so basic, that we really cannot speak about it as an ontology. However, the semantic perspective approach recognizes that different entities will observe concepts from different sets of perspectives (aka, different types). These types however (sets of patterns) do not define reality, as we have seen earlier, they rather observe reality in a particular way. An entity is free to have its own particular sets of patterns (types), independent from or without affecting the greater common reality. |
| Ontologies are governed by an owner. The general accepted trend in the current semantic world in computation is that ontologies are owned by the entities that author and maintain them. This gives the entity absolute power over the reality its ontology generates. Should the entity decide to change the ontology, this has widespread consequences over the realities they spawn. | Nobody owns reality, owners only own their particular view of reality. Ontologies do not define the reality, the common reality exists outside of any ontology. Ontologies only state what an entity’s particular view of it is, and the entity is free to own that, change it, discard it, do whatever they like with it, without impacting the common reality. |
| Semantic web emerges from the unification of all ontologies or supremacy of one. Semantics in the semantic web lies in what computers can do with the reality they have access too (the materialization of the ontology they are programmed to work with). As such, a true semantic web (something akin to the current world wide web) can only emerge from the unification of all the disparate ontologies that exist out there or if one ontology overpowers all others and achieves world wide acceptance. Otherwise we are talking really of many small semantic webs. | Semantic web emerges by majority consensus. Although people are free to have their own perspective on everything, as humans, we also have many things in common. Communication, knowledge exchange relies on these common patterns that we share. The level of similarity between our vantage point is dictated by the degree of overlapping of our interest. We still retain our particular perspectives, but we are driven by interest to form common ones as well.This kind of flexibility exists within the semantic perspective world too. SPInDL does allow all entities to have singular knowledge on everything, but at the same time it also allows that information to be processed under a different set of types that is common with other entities. The semantic perspective assumes that only interest that is common is worth exchanging, the rest is irrelevant. This can only be true if the ontologies do not generate reciprocally exclusive realities, in other words, if a reality allows for multiple ontologies, or if reality defines ontologies and not vice versa. |
| Ontology based realities are fully accountable. The rigorous a-pirori structure imposed by ontologies ensure that whatever reality is created based on them, will strictly abide all the rules defined in the ontology. Ontologies are strict, reliable and in line with the nature of programming languages. | Here be dragons. Semantic Perspective is the world of subjectivity, infinite possibilities, incomplete information, patterns, paradoxes and other pleasantries that lie within the realm of “Information”. It is a world of pattern recognition, a world without structure, without rigor and very few rules that allows for openness and continuous unobstructed evolution. This is something our brains are very comfortable with, but something that might present a great challenge in a computing world that is build on structure and rigor. Unfortunately I do not think we have a choice. We want to represent “information”, and this is the nature of “information”, whether we like it or not. |
After all this has been said, it stands clear that the two approaches could not be any more different than they are. The standard approach is a constructive, synthetic approach. It creates things, it creates reality. The Semantic Perspective model on the other hand is exactly opposite in its approach. It is deconstructive, analytic in nature, it does not construct any reality, only subjective observations of the one and only.
For a deeper understanding of the model I welcome you download the manuscript to explore the detailed philosophy that stands behind the model.