Release 0.1.7

• add procedural var intro
• add some vision modules

Release 0.1.5

various small fixes

procedural

• evidence depends on op, gets rid of global evidence hack which was necessary to play TTT effectively

Release 0.1.4

• resource distribution is now max utility based in procedural reasoner, system decides on runtime how much CPU to spend for some derivations

Release 0.1.2

small updates which make it into a stable version

implementation:

• changed SentenceDummy to Sentence (name is now stable API)

Release 0.1.0

bumping higher release because it seems to be stable and capable enough to do it now before more experimental changes are done again

Release 0.0.30

• switched attention to better attention heavily inspired by ONA

Release 0.0.26

• ExtInt & is supported

Release 0.0.24

• added automatic evaluation of Q&A
• eval: added some examples for evaluation
• Q&A is a bit better

Release 0.0.22

• fixed deadlock
• complexity of declarative judgements is capped to put under holy AIKR

known bugs:

• small bug in parser - it can accept partial lines

known problems:

• revision of temporal beliefs isn't done under AIKR

Release 0.0.20

stable because Pong3 and TTT work

evaluation result:

TTT:
w/l ratio = 2.1875 games = 500
@ max # goals = 1000

known bugs:

• the declarative reasoner causes a deadlock, because the mutexes for the concepts are locked incorrectly
• small bug in parser - it can accept partial lines

known problems:

• term complexity isn't checked before putting beliefs into the system, which puts the system out of AIKR if the outside system feeds it with extremely complicated terms or if the reasoner uses derivations, which produce more complex conclusions (there is only one rule which isn't used in the NLP example(s))
• revision of temporal beliefs isn't done under AIKR