JeTheme | tig12.net

This software is organized around a storage containing data coming from real life : persons, events, groups, links. Data can be represented using different "approaches", like astrology or numerology. Individual and collective representations are available.

I mainly used jetheme to build my genealogical tree, which is a merge between information I wrote in structured yaml files, gedcom files coming from family members and information retrieved by program on genealogical sites (16 pieces of my genealogical tree merged with the information I typed).

Then I implemented chi square tests on Gauquelin data from cura.free.fr. I wrote the whole chain : define a group, pre-compute all useful informations (astrological and numerological data of the persons of the group), compute expected and observed repartitions, display repartitions, compute chi square and p-values. I observed planetary effects everywhere ! Meaningless because expected repartitions were not correctly implemented : I started to implement Gauquelin 1957 book for astronomical corrections using what Ertel calls method n°4 (Gauquelin's standard method) in the book "The tenacious mars effect" (Ertel / Irving, 1996), p SE - 39. Then realized that I would need supplementary information to compute demographical deviations. I decided to generate expected repartitions using shuffling (called methods 5 and 6 by Ertel), and didn't do it because time was missing.

I then realized that the first thing to do is to gather data and started an other program, Gauquelin5.
Jetheme is installed on my machine, I use it occasionaly to work on my genealogical tree ; I haven't worked on it since 2017, but I'll be back on this software.

I have not published jetheme code because it's really at a draft stage.
Main flaws of the program :

The storage. I'm stuck with graph databases because there is no standard like SQL for relational databases. SPARQL from W3C, Cypher from neo4j, Cayley from Google... the Gremlins API ? I didn't know which solution to adopt. Then I used a relational database (postgresql) to simulate a graph database structure, using YAGO model (s, p, o, t, l, x) - subject predicate object time location context. I found that really great : I have only one type of table to store all type of data : person, parent-child relation, person-group relation etc. Great because I can add new types of things in the storage without changing the structure of the database. The inconvenient is that I must implement myself the navigation through the graph. Current implementation works, using cache and pre-computations for complex queries, but it's not designed to scale to large datasets, and my purpose is not to implement graph dtabase on top of postgresql !
The database structure is too complex : each user of the application has one postgresql cluster, to guarantee a separation of data and secure the privacy. Convenient because a user have at his/her disposal several databases, each composed by several schemas. But too heavy to manage.
The program can import gedcom files and yaml files written in a syntax called jth1. But exporting a tree managed by jetheme to a standard format (gedcom) has not been coded. This is a major inconvenient because it makes users prisonner from jetheme.

The web application itself has a basic MVC structure, coded in php.