Pytholog as a logic database

Overview

We will use DVD Rental database to feed a knowledge base as facts and rules, then logically query the database.

Here we can find how to create the database in postgresql and insert the data.

Let's connect to the database in python and see how it looks like:

import psycopg2
import pandas as pd

psql = psycopg2.connect(host = "localhost", database = "dvdrental",
                      user = "postgres", password = "password")
cursor = psql.cursor()
## fetch some data to confirm connection
pd.read_sql("SELECT * FROM language;", psql)

#    language_id                  name         last_update
# 0            1  English              2006-02-15 10:02:19
# 1            2  Italian              2006-02-15 10:02:19
# 2            3  Japanese             2006-02-15 10:02:19
# 3            4  Mandarin             2006-02-15 10:02:19
# 4            5  French               2006-02-15 10:02:19
# 5            6  German               2006-02-15 10:02:19

Let's see what the table names are:

cursor.execute("select relname from pg_class where relkind='r' and relname !~ '^(pg_|sql_)';")
print(cursor.fetchall())

# [('actor',), ('store',), ('address',), ('category',), ('city',), ('country',), 
# ('customer',), ('film_actor',), ('film_category',), ('inventory',), ('language',), 
# ('rental',), ('staff',), ('payment',), ('film',), ('movies_rental',), ('compressed_movies_rental',)]
def query_defn(table):
    return f"SELECT * FROM {table};"

Reading tables into python

No we will read the tables we will query into python and do some transformation to have values in lowercase.

actor = pd.read_sql(query_defn("actor"), psql)
actor["Actor"] = (actor["first_name"] + "_" + actor["last_name"]).str.lower()
actor.head()

#    actor_id first_name  ...             last_update                Actor
# 0         1   Penelope  ... 2013-05-26 14:47:57.620     penelope_guiness
# 1         2       Nick  ... 2013-05-26 14:47:57.620        nick_wahlberg
# 2         3         Ed  ... 2013-05-26 14:47:57.620             ed_chase
# 3         4   Jennifer  ... 2013-05-26 14:47:57.620       jennifer_davis
# 4         5     Johnny  ... 2013-05-26 14:47:57.620  johnny_lollobrigida
# [5 rows x 5 columns]
language = pd.read_sql(query_defn("language"), psql)
film = pd.read_sql(query_defn("film"), psql)
category = pd.read_sql(query_defn("category"), psql)
#customer = pd.read_sql(query_defn("customer"), psql)
language["name"] = language["name"].str.lower()
film["title"] = film["title"].str.replace(" ", "_").str.lower()
category["name"] = category["name"].str.lower()
#customer["Customer"] = (customer["first_name"] + "_" + customer["last_name"]).str.lower()
film_category = pd.read_sql(query_defn("film_category"), psql)
#film[film.film_id.isin(film_category[film_category.category_id == 14].film_id)]

print(film.loc[film.film_id == 1, "title"])
# 4    academy_dinosaur
# Name: title, dtype: object

print(actor.head())
#    actor_id first_name  ...             last_update                Actor
# 0         1   Penelope  ... 2013-05-26 14:47:57.620     penelope_guiness
# 1         2       Nick  ... 2013-05-26 14:47:57.620        nick_wahlberg
# 2         3         Ed  ... 2013-05-26 14:47:57.620             ed_chase
# 3         4   Jennifer  ... 2013-05-26 14:47:57.620       jennifer_davis
# 4         5     Johnny  ... 2013-05-26 14:47:57.620  johnny_lollobrigida

Defining Knowledge Base

Let's initiate the knowledge base and feed it with for loops.

We will use rules as the query statements and views if we need some joining and conditions.

import pytholog as pl
dvd = pl.KnowledgeBase("dvd_rental")

for i in range(film.shape[0]):
    dvd([f"film({film.film_id[i]}, {film.title[i]}, {film.language_id[i]})"])

for i in range(language.shape[0]):
    dvd([f"language({language.language_id[i]}, {language.name[i]})"])

## simple query 
dvd(["film_language(F, L) :- film(_, F, LID), language(LID, L)"])
dvd.query(pl.Expr("film_language(young_language, L)"))
# [{'L': 'english'}]

Rules as views

film_category

We will create film_category view

for i in range(category.shape[0]):
    dvd([f"category({category.category_id[i]}, {category.name[i]})"])

for i in range(film_category.shape[0]):
    dvd([f"filmcategory({film_category.film_id[i]}, {film_category.category_id[i]})"])

dvd(["film_category(F, C) :- film(FID, F, _), filmcategory(FID, CID), category(CID, C)"]) ## "_" to neglect this term

## another query to see what films in sci-fi category 
dvd.query(pl.Expr("film_category(F, sci-fi)"))

# [{'F': 'annie_identity'},
#  {'F': 'armageddon_lost'},
# .....

#  {'F': 'titans_jerk'},
#  {'F': 'trojan_tomorrow'},
#  {'F': 'unforgiven_zoolander'},
#  {'F': 'vacation_boondock'},
#  {'F': 'weekend_personal'},
#  {'F': 'whisperer_giant'},
#  {'F': 'wonderland_christmas'}]

film_actor

Let's join actors and films

for i in range(actor.shape[0]):
    dvd([f"actor({actor.actor_id[i]}, {actor.Actor[i]})"])

film_actor = pd.read_sql(query_defn("film_actor"), psql)
#print(film_actor[film_actor["actor_id"] == 3].shape)
print(film_actor.shape)
#(5462, 3)
for i in range(film_actor.shape[0]):
    dvd([f"filmactor({film_actor.film_id[i]}, {film_actor.actor_id[i]})"])

dvd(["film_actor(F, A) :- film(FID, F, _), filmactor(FID, AID), actor(AID, A)"])

dvd.query(pl.Expr("film_actor(annie_identity, Actor)"))
#[{'Actor': 'adam_grant'}, {'Actor': 'cate_mcqueen'}, {'Actor': 'greta_keitel'}]

## query actors in a film
dvd.query(pl.Expr("film_actor(academy_dinosaur, Actor)"))
# [{'Actor': 'penelope_guiness'},
#  {'Actor': 'christian_gable'},
#  {'Actor': 'lucille_tracy'},
#  {'Actor': 'sandra_peck'},
#  {'Actor': 'johnny_cage'},
#  {'Actor': 'mena_temple'},
#  {'Actor': 'warren_nolte'},
#  {'Actor': 'oprah_kilmer'},
#  {'Actor': 'rock_dukakis'},
#  {'Actor': 'mary_keitel'}]

### query films that an actor performed in
dvd.query(pl.Expr("film_actor(Film, penelope_guiness)"))
# [{'Film': 'academy_dinosaur'},
#  {'Film': 'anaconda_confessions'},
#  {'Film': 'angels_life'},
#  {'Film': 'bulworth_commandments'},
#  {'Film': 'cheaper_clyde'},
#  {'Film': 'color_philadelphia'},
#  {'Film': 'elephant_trojan'},
#  {'Film': 'gleaming_jawbreaker'},
#  {'Film': 'human_graffiti'},
#  {'Film': 'king_evolution'},
#  {'Film': 'lady_stage'},
#  {'Film': 'language_cowboy'},
#  {'Film': 'mulholland_beast'},
#  {'Film': 'oklahoma_jumanji'},
#  {'Film': 'rules_human'},
#  {'Film': 'splash_gump'},
#  {'Film': 'vertigo_northwest'},
#  {'Film': 'westward_seabiscuit'},
#  {'Film': 'wizard_coldblooded'}]

### simple yes or no query
dvd.query(pl.Expr("film_actor(academy_dinosaur, lucille_tracy)"))
# ['Yes']

actor_category

Actor Category view to see in which categories an actor performed.

dvd(["actor_category(A, C) :- film_actor(F, A), film_category(F, C)"])

jd = dvd.query(pl.Expr("actor_category(jennifer_davis, Category)"))

from pprint import pprint
merged = {}
for d in jd:
    for k, v in d.items():
        if k not in merged: merged[k] = set()
        merged[k].add(v)

pprint(merged)
# {'Category': {'action',
#               'animation',
#               'comedy',
#               'documentary',
#               'drama',
#               'family',
#               'horror',
#               'music',
#               'new',
#               'sci-fi',
#               'sports',
#               'travel'}}

Saving knowledge base to prolog file

Finally, let's now write those facts and rules to a prolog file.

with open("dvd_rental.pl", "w") as f:
    for i in dvd.db.keys():
        for d in dvd.db[i]["facts"]:
            f.write(d.to_string() + "." + "\n")

f.close()