| … | … | @@ -33,8 +33,8 @@ math and code. The terminology will bounce between abstract concepts and |
| 33 | 33 | low-level code. Apologies in advance if it veers off into the opaque. |
| 34 | 34 | The topic is difficult. |
| 35 | 35 | |
| 36 | | Genealogy |
| 37 | | --------- |
| 36 | An Example from Genealogy |
| 37 | ------------------------- |
| 38 | 38 | A concrete example is in order, one that sits part-way between |
| 39 | 39 | abstraction and algorithmic directness. Genealogical research provides |
| 40 | 40 | such an example. |
| … | … | @@ -108,6 +108,37 @@ fact that all data is directly available from the operating system, |
| 108 | 108 | using very pure and well-defineed, structured software interfaces, this |
| 109 | 109 | turns out to be a hard problem. |
| 110 | 110 | |
| 111 | Project Goals |
| 112 | ------------- |
| 113 | The goal of the project is to be able to provide a generic framework for |
| 114 | perceiving structural factoids, and stitching them together. Earlier |
| 115 | work insists that structural factoids resemble jigsaw puzzle pieces, |
| 116 | which can be stitched together to form sheaves. [AtomSpace Sheaves](https://github.com/opencog/atomspace/tree/master/opencog/sheaf) |
| 117 | provides a general introduction; the idea has a long pedigree in |
| 118 | academia, dating back to the 1960's. |
| 119 | |
| 120 | Here, the idea is modified: The jigsaw connector tabs are not strict |
| 121 | yes/no matches, but have some confidence or likelihood. This confidence |
| 122 | or likelylihood is obtained from hierarchical subrelations: do |
| 123 | sub-pieces also fit? This confidence may also be obtained by spreading |
| 124 | over the local neighborhood: is coherence maintained over not just |
| 125 | nearest neighbors, but also next-nearest-neighbors, etc.? In Link |
| 126 | Grammar, this is called a "linkage": all jigsaws must be linkable, from |
| 127 | one end to the other; this is a global property infered from the local |
| 128 | linkage constraints. |
| 129 | |
| 130 | There do structural factoids come from? In natural language parsing, |
| 131 | they come from a "lexis", a dictionary of possible forms created by |
| 132 | professional linguists. That lexis is presumed to encompass the |
| 133 | morpho-syntactic structure of language. The lessons of structural |
| 134 | linguistics and syntactic analysis from the 1990's-2010's is that such |
| 135 | dictionaries are always incomplete, and that no hurculean effort by grad |
| 136 | students can ever come to a conclusion. The task is overwhelming. At the |
| 137 | same time, transformers, such as BERT, and now, more generally DL-NN |
| 138 | LLM's have shown that statistica techniques, taken from corpus |
| 139 | linguistics, are far more adept at encoding the structure of language. |
| 140 | |
| 141 | but also |
| 111 | 142 | |
| 112 | 143 | |
| 113 | 144 | Status |