House Hunting Skill

This skill helps users find the best value rental houses using a circle-based search methodology centered on their office location.

Core Methodology

Search Philosophy

1. Center Point: User's office location (geocoded to coordinates)
2. Radius: User's acceptable commute distance/time
3. Scope: ALL residential communities within the circle
4. Goal: Find the best value housing by comparing multiple factors

Evaluation Factors (综合评估)

Workflow

Step 1: Collect User Requirements

Ask the user for the following information:

1. Office Location: Ask "请告诉我你的办公地点（地址或地标）"
2. Commute Distance: Ask "你可以接受的通勤距离是多少？（单位：公里，或公交/地铁通勤时间：分钟）"
3. Room Type: Ask "你要找什么房型？（如：一室、两室、三室等）"
4. Rent Range: Ask "你的租金范围是多少？（如：2000-4000元/月）"
5. Building Age Requirement: Ask "你对房子年限有什么要求？（如：不超过20年、15年以内等）"

IMPORTANT - Building Age Filter:

• If user specifies "不超过X年" or "X年以内", then completely exclude any community/property built older than X years
• Example: User says "不超过20年" → Only show communities built in 2006年以后（含2006）
• This is a hard filter, not just a scoring factor - do NOT show properties that exceed the age limit

Step 2: Geocode Office Location

Convert the office address to latitude and longitude coordinates using a geocoding service.

Recommended APIs:

• Tencent Map Geocoding: https://apis.map.qq.com/jsapi?address={address}&key={key}
• Amap Geocoding: https://restapi.amap.com/v3/geocode/geo?address={address}&key={key}

Output: {lat, lng} coordinates of the office location

Step 3: Determine Search Radius

Convert user's commute preference to a search radius:

• If user provides distance (km): Use directly as radius
• If user provides time (minutes): Estimate radius based on average transit speed (~15-20 km/h for city transit)

Estimated Radius Conversion:

• 30分钟 transit ≈ 7.5-10 km radius
• 45分钟 transit ≈ 11-15 km radius
• 60分钟 transit ≈ 15-20 km radius

Step 4: Find All Communities Within Circle

Search for ALL residential communities within the specified radius from the office:

1. Use Map APIs to find nearby residential areas:
• Tencent Map nearby search: https://apis.map.qq.com/ws/geocoder/v1/?location={lat},{lng}&poi_options=policy=3
• Or search for residential communities by area

1. Expand search to cover the entire circle:
• Query multiple districts/areas within the radius
• Collect all community names within the geographic scope

1. Cross-reference with rental platforms:
• For each community found, search for rental listings on multiple platforms

Step 5: Search Rental Listings for Each Community

IMPORTANT: For each community found within the circle, search on at least two platforms:

Search Parameters:

• Community name: Search by specific community
• Price range: User's rent range
• Room type: User's preferred room type (2室, 3室)
• Housing type: 整租 (Whole rental only) - exclude 合租, 单间, 床位

Step 6: Calculate Commute Time for Each Listing

For each candidate listing, calculate the public transit commute time from the listing to the office.

Tencent Map Transit API:

https://apis.map.qq.com/direction/transit?from={start_lat},{start_lng}&to={end_lat},{end_lng}&key={key}&policy=LEAST_TIME

Amap Transit API:

https://restapi.amap.com/v3/direction/transit?origin={start_lng},{start_lat}&destination={end_lng},{end_lat}&key={key}&strategy=0

Step 7: Validate Legitimate Residential Communities

CRITICAL: Only include listings from legitimate residential communities.

❌ EXCLUDE Types:

1. 城中村 (Urban Villages): Names containing 村, 巷, 里, 街 (village-style)
2. 研发/工业公寓: Names containing 研发公寓, 工业园
3. 非住宅性质: 农民房, 小产权房, 商住两用

✅ INCLUDE Types:

• 正规住宅小区: Names with 小区, 花园, 苑, 居, 园, 湾, 郡, 府
• 商品房住宅: Regular commercial housing with proper property management

Step 8: Gather Community & Property Details

For each community, collect:

• Building age: Year built (房龄/年限)
• Community environment: 绿化率, 容积率, 配套设施
• Property type: 普通住宅 vs other types
• Elevator availability: 电梯房 vs 楼梯房

Sources:

• 百度百科 (Baidu Baike) for community details
• 房天下 (Fang.com) for property information
• 链家/贝壳 (Lianjia/Beike) for building age

Step 9: Calculate Value Score (性价比评分)

For each listing, calculate a comprehensive value score:

总分 = 通勤得分×30% + 小区环境得分×25% + 房龄得分×20% + 租金性价比得分×25%

Scoring Details:

⚠️ IMPORTANT - Building Age as Hard Filter:

• The user's building age requirement (e.g., "不超过20年") is a HARD LIMIT, not just a scoring factor
• Any community/property older than the limit must be completely excluded, regardless of other qualities
• Example: User requires "不超过20年" (current year 2026 → built after 2006):
• ✅ 桃源村一期(1997) → EXCLUDE (29年 > 20年)
• ✅ 华联花园(1997) → EXCLUDE (29年 > 20年)
• ✅ 彩虹居(1998) → EXCLUDE (28年 > 20年)
• ✅ 荔园新村(1995) → EXCLUDE (31年 > 20年)
• ✅ 松坪村(2005左右) → EXCLUDE (~21年 > 20年)

Step 10: Present Ranked Results

Display listings ranked by comprehensive value score:

Summary:

• Top 3 best value listings with detailed reasoning
• Trade-off analysis (e.g., longer commute for better community)
• Platform cross-reference status

Important Notes

• Circle-based search: Always search by geographic radius from office, not by administrative district
• Comprehensive evaluation: Consider ALL factors (commute, community, age, price) not just price
• Building Age is a HARD FILTER: If user specifies a maximum age (e.g., "不超过20年"), completely exclude any property older than that limit - do NOT show them even if they score high on other factors
• Only 正规小区房: Exclude 城中村, 农民房, 研发公寓
• Only 整租: Exclude 合租, 单间, 床位
• Cross-reference at least 2 platforms: Verify data accuracy
• Prioritize overall value over single factor: A slightly higher rent for much better commute/location may be worth it

Supported Room Types

• 一室 (1-bedroom)
• 两室 (2-bedroom)
• 三室 (3-bedroom)
• 及以上等

Platform URLs for Reference

See references/platforms.md for detailed API patterns.