Prediction-Ready Entity Reconciliation | Dataford Interview Questions

Write up a prediction-ready entity reconciliation system for us on your personal computer.

Implement reconcile_entities(records, threshold). Each record is a dictionary containing record_id, name, email, and phone, where values may be empty strings. Return sorted dictionaries with canonical_id, sorted members, and a confidence score. Normalize text case-insensitively and remove non-alphanumeric characters. Use candidate blocking, weighted comparison, and transitive clustering. Two records may be compared only when they share a normalized email, phone, or first-four-character name block. Their score is 0.5 for equal nonempty emails, 0.3 for equal nonempty phones, and 0.2 * SequenceMatcher name similarity. Join records when the score meets threshold, unless two populated identifiers conflict. The canonical record has the most populated fields, with record_id as the tie-breaker.

Constraints

1 <= len(records) <= 2000
Each record contains record_id, name, email, and phone
Field values are strings and may be empty
0.0 <= threshold <= 1.0
record_id values are unique
A populated email or phone conflicts when both records contain different normalized values

Function Signature

def reconcile_entities(records, threshold):

Your question is Prediction-Ready Entity Reconciliation. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

Write up a prediction-ready entity reconciliation system for us on your personal computer.

Constraints

1 <= len(records) <= 2000
Each record contains record_id, name, email, and phone
Field values are strings and may be empty
0.0 <= threshold <= 1.0
record_id values are unique
A populated email or phone conflicts when both records contain different normalized values

Function Signature

def reconcile_entities(records, threshold):

Your question is Prediction-Ready Entity Reconciliation. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.