Backward and forward chaining algorithm for (expert system) in Python

Question

I'm looking for the algorithm of backward and forward chaining to implement it with Python. I looked on the internet, but I didn't find too much. I looked in Wikipedia too but I just found some rules and I didn't find the algorithms.

Answer 1

I know you originally tagged your earlier question with Python so I will limit this to just Python.

When looking for code examples one good place to look is in the Github repositories. Querying for backward chaining and then limiting the results to Python will yield this query You can do the same for forward chaining.

Note that you will have to find the code samples you like and then thoroughly test it before making using of it. There are great examples there, but more likely attempts at it that are incorrect. Plan to spend a few days with them and create lots of test cases. All of the code there comes with no guarantees.

If you want just the algorithm then find books for Artificial Intelligence such as Artificial Intelligence ... by George F Luger or Artificial Intelligence ... by Russell and Norvig.

Answer 2

Forward-chaining inference engines can be implemented relatively easily in Python. This is a list of inference rules:

mammal(A) ==> vertebrate(A).
vertebrate(A) ==> animal(A).
vertebrate(A),flying(A) ==> bird(A).
vertebrate("duck").
flying("duck").
mammal("cat").

These rules can be translated into Python:

global facts
global is_changed

is_changed = True
facts = [["vertebrate","duck"],["flying","duck"],["mammal","cat"]]

def assert_fact(fact):
    global facts
    global is_changed
    if not fact in facts:
        facts += [fact]
        is_changed = True

while is_changed:
    is_changed = False
    for A1 in facts:
        if A1[0] == "mammal":
            assert_fact(["vertebrate",A1[1]])
        if A1[0] == "vertebrate":
            assert_fact(["animal",A1[1]])
        if A1[0] == "vertebrate" and ["flying",A1[1]] in facts:
            assert_fact(["bird",A1[1]])

print(facts)

From the initial set of facts, the inference engine generates this list:

[['vertebrate', 'duck'], ['flying', 'duck'], ['mammal', 'cat'], ['animal', 'duck'], ['bird', 'duck'], ['vertebrate', 'cat'], ['animal', 'cat']]

Answer 3

global facts
global is_changed

is_changed = True
facts = [["plant","mango"],["eating","mango"],["seed","sprouts"]]

def assert_fact(fact):
    global facts
    global is_changed
    if not fact in facts:
        facts += [fact]
        is_changed = True

while is_changed:
    is_changed = False
    for A1 in facts:
        if A1[0] == "seed":
            assert_fact(["plant",A1[1]])
        if A1[0] == "plant":
            assert_fact(["fruit",A1[1]])
        if A1[0] == "plant" and ["eating",A1[1]] in facts:
            assert_fact(["human",A1[1]])

print(facts)