Shop Naturally - talk and find a new phone

Search Process Overview

The search process involves several steps executed via a MongoDB aggregation pipeline:

1. Filtering ($match):

   • Applies filters based on user input (price range, RAM, storage, brand, rating, stock status, name, reviews, text search terms, popularity).
   • Includes a specific recommendation logic: if only a maximum price is set, it narrows the search to a ₹5000 range below that maximum.

2. Weighted Scoring ($addFields):

   • Calculates a score for each phone to determine its relevance beyond simple filtering.
   • Different features contribute differently to the score (weights):
     • bought: 4x
     • reviews: 3x
     • ratingFloat: 2x
     • ram: 1x
     • storage: 1x
   • Missing values (null) for these fields are treated as 0 during score calculation ($ifNull).

3. Initial Sorting ($sort):

   • Phones are sorted primarily by the calculated score in descending order (highest score first).
   • price (ascending, lowest first) is used as a secondary sort key to break ties.

4. Brand Diversity Enforcement:

   • Group by Brand ($group): Phones are grouped by their brand.
   • Limit per Brand ($project & $slice): Each brand group is limited to its top 4 phones (based on the initial sort).
   • Recombine ($unwind & $replaceRoot): The limited groups are merged back into a single list, ensuring no brand has more than 4 entries.

5. Final Sorting ($sort):

   • The diversified list is sorted again by score (descending) and price (ascending).

6. Result Limiting ($limit):

   • The final list is limited to the top 8 phones.

This multi-stage process aims to return the most relevant phones based on user criteria and popularity/ratings, while also ensuring a variety of brands are presented in the results.