Lesson 15. Ask for clarification instead of guessing

Why this matters

A ticket clerk cannot issue a ticket if the buyer names two films and no showing. They ask which one the buyer wants. Our model must also stop when it cannot tell which class or which kind of information is requested. A refusal here means “not enough verified information for the next step,” not “your question is bad.”

Join the previous steps

Before the code, learn its new marks. text stores the lowercase question, clean removes ? and ,, and words stores its words. entities will hold found club names; intents will hold question types. len(list) counts items. > 1 means “more than one.” [0] takes the first item because Python numbers positions from zero; an empty list has no such item. We therefore count and check before taking an item.

We still do not answer with a time or place: there is no catalog of approved facts yet. іздеу: шахмат уақыт means “we may look for the chess time record,” not “the time was found.” Use the two names from lesson 13 and two question types from lesson 14. Create request.py in UTF-8.

known = ["шахмат", "сурет"]
text = input("Сұрақ: ").lower()
clean = text.replace("?", "").replace(",", "")
words = clean.split()
entities = []
intents = []
for word in words:
    if word in known:
        entities.append(word)
    if word == "қашан":
        intents.append("уақыт")
    if word == "қайда":
        intents.append("орын")
if len(entities) == 0:
    print("білмеймін: қай үйірме?")
elif len(entities) > 1:
    print("нақтылаңыз: бір үйірмені атаңыз")
elif len(intents) == 0:
    print("білмеймін: уақыт па, орын ба?")
elif len(intents) > 1:
    print("нақтылаңыз: бір сұрақты таңдаңыз")
else:
    print("іздеу:", entities[0], intents[0])

The step file matches this example.

len(entities) counts list items; == 0 checks for none. > means “greater than,” so > 1 detects more than one finding. elif branches are tried in order and only the first matching one runs. entities[0] takes the first item; Python counts list positions from zero. We reach this line only after confirming there is exactly one item in both lists. intents[0] works the same way. Taking [0] from an empty list would cause an error; the preceding checks prevent that. == compares while a single = stores a value, as in lesson 10.

Шахмат қашан? gives іздеу: шахмат уақыт. Робот қашан? gives білмеймін: қай үйірме? because that name is not yet in the dictionary. Шахмат, сурет қайда? asks for one club. Шахмат қашан қайда? asks for one question type. Шахмат неге? gives білмеймін: уақыт па, орын ба?. Repeating the same club name also counts as two findings; we will improve that behavior later instead of silently deleting the user’s words.

Memory map

Question → known names + known types → exactly one of each? → lookup key; otherwise → say what is missing or needs clarification. Lookup key → facts (from lesson 16), not an answer yet.

Recall and task

Hide the code and recover the four reasons to stop. Run the examples above and match each message to its reason.

Task: predict the results for СУРЕТ қайда?, Сурет қашан қайда?, and an empty line. Hint: count both lists before checking conditions. Answer: іздеу: сурет орын; нақтылаңыз: бір сұрақты таңдаңыз; білмеймін: қай үйірме?. Common mistake: treating іздеу: as an answer about a location. The model has only identified two keys for a future search; it has not verified a fact.