Wikirace Solver

The search problem

Wikirace asks a player to reach a target Wikipedia article from a starting article by following links. This project models each article as a node and each article link as an edge, then searches for a route between the selected pages.

Breadth-first and iterative deepening

The Go backend provides breadth-first search (BFS) and iterative deepening search (IDS). BFS processes a frontier one depth at a time and tracks visited pages, while IDS repeats depth-limited searches with gradually increasing limits. Both approaches scrape article links rather than relying on a Wikipedia API for graph traversal, so network latency, request limits, and the number of outgoing links all affect the search.

Choose start and target pages
            |
            v
     Select BFS or IDS
            |
            v
Scrape article links and search
            |
            v
Return a route and search metrics
            |
            v
Display the path in the React interface

Frontend and backend

The React frontend accepts the two page titles, lets the user choose BFS or IDS, and draws the returned route as a graph. The Go backend uses Gin for its HTTP API and Colly for scraping pages, with separate search packages for the algorithms. The repositories are split between the frontend and backend.

The project helped me see how a graph-search algorithm behaves when the graph must be discovered through live requests. That made the cost of obtaining neighbors as important as the search strategy itself.