Roadmap » History » Version 9

Jeremy Gow, 2012-11-07 01:49 PM

1 1 Jeremy Gow
h1. Roadmap
2 1 Jeremy Gow
3 9 Jeremy Gow
The development code has been made compatible with sbcl 1.1.
4 9 Jeremy Gow
The immediate goal is to release a version that works with the built-in
5 8 Jeremy Gow
examples (Conklin 95 etc.).  Need to fix problems created by datasets with
6 1 Jeremy Gow
incomplete viewpoints.
7 8 Jeremy Gow
8 9 Jeremy Gow
9 1 Jeremy Gow
h2. Short-term
10 1 Jeremy Gow
11 7 Jeremy Gow
New basic viewpoints:
12 8 Jeremy Gow
* cents - a higher resolution representation of pitch. [High priority for Makam data.]
13 1 Jeremy Gow
* comma (implemented?)
14 1 Jeremy Gow
* metrical contour
15 1 Jeremy Gow
16 9 Jeremy Gow
Make system more data agnostic:
17 9 Jeremy Gow
* Remove dependancy on amuse.
18 9 Jeremy Gow
* Separation of music viewpoints from model.
19 9 Jeremy Gow
* A straightforward interface for specifying viewpoints.
20 1 Jeremy Gow
21 5 Jeremy Gow
Viewpoint selection:
22 5 Jeremy Gow
* Adding viewpoint weights to output.
23 5 Jeremy Gow
* Optionally specify: start point for search, min-links
24 1 Jeremy Gow
* More flexible way for user to specify constraints on viewpoint search:
25 1 Jeremy Gow
** Define labelled viewpoint classes
26 1 Jeremy Gow
** Pairs/triples of labels/wildcards specify acceptable combinations
27 1 Jeremy Gow
** User provides whitelist or blacklist spec
28 1 Jeremy Gow
29 1 Jeremy Gow
Efficiency:
30 6 Jeremy Gow
* Check/extend caching of models etc.
31 6 Jeremy Gow
* Use sampling to estimate mean IC during VP selection.
32 7 Jeremy Gow
33 9 Jeremy Gow
Some basic benchmarks to ensure stability of future development versions.
34 8 Jeremy Gow
35 9 Jeremy Gow
36 1 Jeremy Gow
h2. Mid-term
37 1 Jeremy Gow
38 1 Jeremy Gow
Viewpoint selection:
39 4 Jeremy Gow
* Optimise based on match with existing IC values.
40 5 Jeremy Gow
41 5 Jeremy Gow
Allow user to specify structure of model.
42 4 Jeremy Gow
* Determine order in which distributions are combined.
43 5 Jeremy Gow
* Specify weights for particular combinations, e.g. weighted viewpoints, or weighted memory stores.
44 5 Jeremy Gow
* Multiple memory stores.
45 5 Jeremy Gow
* Specify alternative context strategies.
46 5 Jeremy Gow
* Provide some prepackaged models, e.g. the current model structure.
47 5 Jeremy Gow
48 5 Jeremy Gow
Allow models to use predictive information (PI), expected PI and PI rate (as analogs to IC, entropy and entropy rate respectively).
49 1 Jeremy Gow
50 1 Jeremy Gow
Predict more than one basic viewpoint
51 1 Jeremy Gow
52 1 Jeremy Gow
h2. Long-term
53 1 Jeremy Gow
54 1 Jeremy Gow
Hierarchical structure.
55 7 Jeremy Gow
Parallel implementation.
56 9 Jeremy Gow
57 9 Jeremy Gow
h2. Minor improvements
58 9 Jeremy Gow
59 9 Jeremy Gow
* Show all viewpoints in describe-dataset.
60 9 Jeremy Gow
* Zero barlength (e.g. in Makam data) causes divide by zero error.
61 9 Jeremy Gow
* Update kern import to handle new pitch viewpoints.