Visualizzazione post con etichetta machine learning. Mostra tutti i post
Visualizzazione post con etichetta machine learning. Mostra tutti i post

sabato 3 aprile 2021

PYTHON - Building a memory based on external stimuli

This script parses the incoming phrase with the keys contained in the json file, if it finds the key it processes a response, otherwise it adds the key and rewrites the json file.
import json
#Da stimoli esterni il software allarga le proprie conoscenze

def caricaCoscienza(nomeCoscienza):
    with open(nomeCoscienza,"r") as json_file:
        dictFromJSon=json.load(json_file)
    return dictFromJSon

def memorizza(valori,dict,nomeCoscienza):
    chiave=valori[0]
    lista=[]
    for i in range(len(valori)):
        if i>0:
            lista.append(valori[i])
    dict[chiave]=lista
    with open(nomeCoscienza,"w") as json_file:
            json.dump(dict, json_file)
    with open(nomeCoscienza,"r") as json_file:
        dictFromJSon=json.load(json_file)
    return dictFromJSon

def elabora(valori,ch,keys,dict):
    chiave=keys[ch]
    risposta=dict[chiave]
    return risposta

def ricerca(ingresso,dict,nomeCoscienza):
    keys=list(dict.keys())
    valori=ingresso.split(' ')
    trovato=False
    r=''
    for i in range(len(keys)):
        for c in range(len(valori)):
            if keys[i]==valori[c]:
                trovato=True
                ch=i
    if not(trovato):
        dict=memorizza(valori,dict,nomeCoscienza)
    else:
        r=elabora(valori,ch,keys,dict)
    return r


if __name__ == "__main__":
    dict=caricaCoscienza("result.json")
    dato=input(">")
    print(ricerca(dato,dict,"result.json"))

lunedì 31 agosto 2020

PYTHON - MACHINE LEARNING

I couldn't rely on existing databases for my machine learning project. I need a highly plastic database, so I decided to implement mine. 
Using json and python files I can modify the structures of the tables without having to worry about modifying existing queries, they adapt to changes automatically.
The idea is that, when the machine does not find the data in its databases, it searches for them with the means at its disposal, automatically evaluating which is the best way to search for a given data. The decision tree will automatically expand with the accumulated experience. 
It is not impossible, but it will be long. For now I start building my database.