Class 7 · CBSE AI · Strand C — NLP, Vision, and LLMs Deep-Dive
What is sentiment analysis? Building one — for Class 7
How AI decides if text is positive, negative or neutral — one of the most-used NLP tasks. For Class 7.
What this concept actually says
- Sentiment analysis classifies text as positive, negative, or neutral based on language patterns
- Training data labels determine what 'positive' and 'negative' mean — and those labels embed human biases
- Edge cases like sarcasm, cultural expressions, and code-switching break naive sentiment models
An analogy your child will recognise
Exam paper correction
Imagine grading essays by counting how many times students wrote 'good' or 'bad' — you'd completely miss the student who wrote 'Not bad at all!' meaning they loved it, or the one who wrote sarcastically 'Oh, absolutely brilliant.' A good teacher reads the whole meaning. Sentiment analysis has the same challenge.
Monsoon season — complaints vs. appreciation
In India, someone saying 'The monsoon has really come down on us this year' could be a joyful farmer or a flooded shopkeeper. The same words carry opposite sentiments depending on who's speaking. Teaching an AI which is which requires knowing who the speaker is — not just what they said.
Common misconceptions to watch for
- More positive words than negative words in a sentence always means positive sentiment — negation, sarcasm, and intensifiers can completely invert this.
- A sentiment analyser trained on English product reviews will work just as well on social media posts in regional languages — domain and language mismatch severely degrades performance.
Key facts in one breath
- Sentiment analysis is one of the most widely deployed NLP tasks — used in product reviews, stock market prediction, and political monitoring.
- Sarcasm and irony are the hardest linguistic features for sentiment models to handle, with even state-of-the-art models struggling.
- Negation is a classic failure point: 'not bad', 'not at all satisfied', and 'hardly disappointing' all flip the sentiment of adjacent words.
- Labelling training data for sentiment is subjective — studies show human annotators agree only about 70–80% of the time even on seemingly clear cases.
How Dhee Learning teaches this — the 3-stage question loop
Every Dhee Learning session for this concept follows three stages. We share the questions Dhee actually asks, so you can hear what a session sounds like.
Stage 1 — Surface
A restaurant app wants to automatically sort customer reviews into 'happy' and 'unhappy'. Before building anything, what are three things that could go wrong if you just taught the AI 'good words = positive, bad words = negative'?
Rote answer
"Sentiment analysis checks if a review is positive or negative."
Understood
"Someone could write 'Oh wow, the food took only two hours to arrive — amazing!' sarcastically. The words 'amazing' and 'wow' are technically positive but the meaning is negative. A word-list approach would get this completely wrong."
Stage 2 — Reasoning
If your sentiment analyser was trained mostly on English reviews from Mumbai, why might it perform poorly on reviews written in Hinglish from Jaipur — even if both are about the same restaurant chain?
Follow-up Dhee may use: How would you fix this — would you collect more data, change the model, or something else entirely?
Stage 3 — Application
You're asked to build a sentiment tool that monitors social media posts about a school's new canteen menu. Design the three biggest decisions you'd need to make before writing any code.
Misconception Dhee watches for: Jumping straight to 'use an existing sentiment API' without questioning whether that API's training data matches school-student language and context.
Related concepts
Want your child to actually understand this?
Dhee turns this concept into a 15-minute spoken session — asking, listening, and probing — so your child builds the idea themselves.
Frequently asked questions
What is building a sentiment analyser — explained for kids? +
How AI decides if text is positive, negative or neutral — one of the most-used NLP tasks. For Class 7.
What's the most common mistake children make about this concept? +
More positive words than negative words in a sentence always means positive sentiment — negation, sarcasm, and intensifiers can completely invert this.
How does Dhee Learning teach this in a Class 7 session? +
Dhee opens with a question — for example: "A restaurant app wants to automatically sort customer reviews into 'happy' and 'unhappy'. Before building anything, what are three things that could go wrong if you just taught the AI 'good words = positive, bad words = negative'?" — listens to your child's answer, then probes the reasoning behind it. The session ends when the child can apply the idea to a brand-new situation, not just recall it.