Superimposing Silences

Progress has been made, thanks to the combined efforts of Robert Green, Tommy Cliff, Lin Zou, Professor Hongji Yang and myself. The starting point was the question raised by Robert: what happens if one silence is superimposed on another? In other words, if we combine a silence that has a joyful affect with one that has a tense affect, what is the result?

Given that every silence is a new silence, we may be certain that the result will not be simply a combination of the two previous silences. The affect will be unknowable. However, we may nevertheless undertake some analysis to get some idea of the result.

Mathematically, this leads to set theory, in the first instance, and possibly to group theory in the long run (although that may be overkill). Robert suggests:

The Space (S) of all Silences is partitioned into L disjoint sets of silences (where L is infinite) S1, S2, Sn… Su is a (hopefully empty!) set of silences unclassified by a particular method of partitioning – i.e. the remnants.

We can refine this partitioning by affect and effect, as follows: Sa1, Sa2, Sa3 etc. and Se1, Se2, Se3 etc. It is possible that certain sets have no common elements (silences) then another from the other partition e.g Sa2 ∩ Se3 = ∅. This introduces the idea of incompatibility, i.e. that there silences that cannot coexist at the same time.

When compatible silences are superimposed, therefore, a third silence will be the result, which itself may be judged under the variables in the ontology. This process may be written as follows:

∫:SxS→S, where S is the space of silences. S is two dimensional (affect/effect). The axiom:

  • I. Closure – every product of superimposition is another silence.
  • II. Associativity – the order of superimposition is not relevant and produces the same product.
  • III. Identity – There is an element such that superimposition by it makes no change.
  • IV. Invertibility – that every silence could be superimposed with another to yield the identity.

Robert went on to speculate about how this relates to group theory. If I-IV all hold, it forms a group. If I and II hold, it forms a semi-group. If I, II and III hold, it forms a monoid.

Tommy Cliff commented further that: I. is a question whether two silences combined make a silence. For III, the identity element is ideal silence as well, since if you superimposed a Silence with the identity, you would want to get exactly your original Silence out. I cannot think how you could have the invertibility property (IV) though, ie. for any silence, there exists another silence such that when superimposed, you get ideal silence as your output. If you can’t have this, then you only have a monoid.

Clearly this all  has some way to go, and we settled on using more empirical tools to analyse the silences for now until we can arrive at a judgment as to how much of this is purely subjective speculation (most of it, one suspects).

Hongji Yang proposed using Fourier transforms to analyse the silences. This has the great advantage that it will work across any media type, providing a scientific method for structuring the database. Lin You has undertaken a survey of databases. We have to go for a free platform (there is no budget for this project yet!) so it is a choice between NoSQL and MySQL solutions. Having something that works effectively with semantic web queries is essential. However, this also led on to a discussion about the extent of the use of semantic web in the project. As Jim Hendler puts it: “a little semantics goes a long way”. This is to be resolved. In the meantime, we are absorbing Lin’s survey to make some decisions.

At the same time, the psychological tests being undertaken by Dr Marie Thomas and her team could include superimposed silences. Perhaps something will emerge from that process that gives us an objective measure.

I stressed that user interaction will be key to this project. Whatever decisions we may make about the effect/affect of a given silence, users may well have a different opinion. They will be given the opportunity to record these opinions and the app will learn from their interactions. I envisage something quite alchemical: combine a drop of this silence with a dollop of that one, to produce something quite extraordinary. It could be magical!

Finally, I note that we have already had some uploads of silences. The individuals are beginning to appear! It’s exciting.