Neo4j & Graph Databases — Cypher

9. Practice Problems

All problems use the social graph from §4 (Arif, Mim, Tanvir, Nadia, Rakib + Grameenphone, bKash). For each, write or read a Cypher query. Click Show Answer after attempting.

1. Write a Cypher query to return the name and city of every :Person node, sorted by name.
   প্রতিটি :Person node-এর নাম ও শহর name অনুযায়ী sorted দেখান।
   ✨ Show Answer (উত্তর দেখুন)

   ans1.cypher

   MATCH (p:Person)
   RETURN p.name AS name, p.city AS city
   ORDER BY name;

2. Find the direct friends (1 hop, outgoing :KNOWS) of 'Mim'.
   'Mim'-এর সরাসরি বন্ধুদের তালিকা।
   ✨ Show Answer

   ans2.cypher

   MATCH (:Person {name:'Mim'})-[:KNOWS]->(f:Person)
   RETURN f.name AS friend
   ORDER BY friend;

3. For every person, count how many outgoing :KNOWS relationships they have. Include people with 0.
   প্রতিটি Person-এর outgoing :KNOWS সংখ্যা গণনা করুন (০-ও দেখান)।
   ✨ Show Answer

   ans3.cypher

   MATCH (p:Person)
   OPTIONAL MATCH (p)-[:KNOWS]->(f:Person)
   RETURN p.name AS person, count(f) AS friends_out
   ORDER BY friends_out DESC;

4. Find every person reachable from Arif within exactly 2 KNOWS hops.
   Arif থেকে ঠিক 2 hop দূরে আছেন এমন person খুঁজে বের করুন।
   ✨ Show Answer

   ans4.cypher

   MATCH (:Person {name:'Arif'})-[:KNOWS*2]->(p:Person)
   RETURN DISTINCT p.name AS two_hops_away
   ORDER BY two_hops_away;

5. Use shortestPath to find the shortest :KNOWS chain from Arif to Rakib.
   Arif থেকে Rakib পর্যন্ত সবচেয়ে ছোট :KNOWS path বের করুন।
   ✨ Show Answer

   ans5.cypher

   MATCH path = shortestPath(
       (:Person {name:'Arif'})-[:KNOWS*]-(:Person {name:'Rakib'})
   )
   RETURN [n IN nodes(path) | n.name] AS chain,
          length(path) AS hops;

6. Suggest friends for Arif: people he doesn't know, but his friends do, ranked by mutual friend count.
   Arif-কে friend suggestion দিন — যাদের সে চেনে না কিন্তু তার বন্ধুরা চেনে।
   ✨ Show Answer

   ans6.cypher

   MATCH (me:Person {name:'Arif'})-[:KNOWS]->(f)-[:KNOWS]->(fof)
   WHERE NOT (me)-[:KNOWS]->(fof) AND me <> fof
   RETURN fof.name AS suggestion,
          count(DISTINCT f) AS mutual_friends
   ORDER BY mutual_friends DESC;

7. Find every pair of :Person nodes who work at the same company.
   কোন দুজন একই কোম্পানিতে কাজ করেন — তাদের জোড়া দেখান।
   ✨ Show Answer

   ans7.cypher

   MATCH (a:Person)-[:WORKS_AT]->(c:Company)<-[:WORKS_AT]-(b:Person)
   WHERE id(a) < id(b)
   RETURN a.name AS p1, b.name AS p2, c.name AS company;

   আমরা id(a) < id(b) দিয়ে প্রতিটি জোড়া কেবল একবার দেখাচ্ছি।

8. Use MERGE to add a new :KNOWS relationship from Tanvir to Nadia, but do not duplicate if it already exists.
   Tanvir থেকে Nadia-তে :KNOWS যোগ করুন, কিন্তু আগে থাকলে duplicate করবেন না।
   ✨ Show Answer

   ans8.cypher

   MATCH (t:Person {name:'Tanvir'}), (n:Person {name:'Nadia'})
   MERGE (t)-[r:KNOWS]->(n)
   ON CREATE SET r.since = 2024
   RETURN r;

9. Create a uniqueness constraint on Person.email.
   Person.email-এর উপর uniqueness constraint তৈরি করুন।
   ✨ Show Answer

   ans9.cypher

   CREATE CONSTRAINT person_email_unique
   FOR (p:Person) REQUIRE p.email IS UNIQUE;

10. In one sentence, when does a graph database beat a relational database?
    এক বাক্যে বলুন — graph database কখন SQL-এর চেয়ে ভালো?
    ✨ Show Answer

    Answer: When the queries you ask depend on traversing many relationships (2+ hops, recommendations, ring detection, permission inheritance), a graph database stores those connections as direct pointers and walks them in constant cost per hop, while SQL would need a JOIN per hop and slow down dramatically with depth.

    যখন query-এর উত্তর পেতে অনেক hop traversal দরকার (২+ hop, recommendation, fraud ring, permission inheritance) — তখন graph database সরাসরি pointer follow করে, SQL-এর JOIN-এর তুলনায় অনেক দ্রুত।

11. Why is variable-length matching like [:KNOWS*] dangerous in production?
    [:KNOWS*] production-এ কেন বিপজ্জনক?
    ✨ Show Answer

    Answer: Without an upper bound, the engine may try to walk paths of arbitrary length, which on a dense graph can produce billions of intermediate paths and exhaust memory. Always pin an explicit upper limit such as *1..4, or use shortestPath.

    উপর-সীমা না থাকলে engine অগণিত path generate করতে পারে, dense graph-এ এটি memory শেষ করে দেবে। সবসময় *1..4-এর মতো explicit সীমা দিন।

12. Describe one real fraud-detection pattern that a graph database can match easily.
    একটি বাস্তব fraud-detection pattern বর্ণনা করুন যেটি graph-এ সহজে match করা যায়।
    ✨ Show Answer

    Answer: A "fraud ring" — multiple seemingly-unrelated accounts that share an underlying identifier such as a device, IP address, or recovery phone number. In Cypher: MATCH (a1:Account)-[:USES]->(d:Device)<-[:USES]-(a2:Account) WHERE a1 <> a2 RETURN ... — instantly returns every pair of accounts sharing a device. Add more nodes (IP, card BIN, address) to detect bigger rings.

    "Fraud ring" — আপাতদৃষ্টিতে আলাদা একাধিক account যারা একটি device, IP বা phone number শেয়ার করে। graph-এ এক pattern match-ই সবগুলো ring বের করে দেয়।