Loading...
A search engine separates discovering pages, organising their contents and ranking results. PageRank illustrates how a link graph can provide an importance signal without reading each link as an equal vote.
What you'll be able to do
A crawler discovers and fetches resources, often following links from known pages. The engine analyses suitable fetched content and builds an index connecting search terms or other features to relevant documents. An inverted index maps a term to postings for documents containing it, avoiding a complete scan of every stored page for each ordinary query.
For pages P1="robot club" and P2="robot design", the posting list for robot includes both pages, while design includes P2. Real indexes may record positions, frequencies and other information. A page being discovered does not guarantee it is fetched, indexed or highly ranked; content can also change after an index was built.
In the PageRank model, a page receives contributions from pages linking to it. A contribution depends on the linking page’s own rank and how that rank is divided across its outgoing links. Thus a link from an important page can count differently from a link from a less important page. The ranks are interdependent and are refined iteratively.
For an original one-round illustration, let A have rank 0.6 and two outgoing links, to B and C. Before damping, A contributes 0.3 to each. Let B have rank 0.3 and one outgoing link to C, contributing another 0.3. These contributions are not a full converged ranking calculation: they isolate why both source rank and out-degree matter.
The random-surfer interpretation includes following links and sometimes jumping to another page. Damping reduces the problem of rank becoming trapped in a closed link group. With three pages, damping 0.85 and equal starting jump allocation, each page has a base term (1−0.85)/3=0.05. A page’s updated value adds 0.85 times its incoming contributions in this simplified step.
OCR requires a high-level understanding, not implementing the algorithm. Link importance is also different from relevance to a particular query: a highly linked page about music need not answer a robotics question. Search systems can combine many signals and address spam or manipulated links. Do not describe this historical model as the sole rule used by every current search engine.
| Source | Assumed rank | Outgoing targets | Contribution before damping |
|---|---|---|---|
| A | 0.6 | B and C | 0.3 to each |
| B | 0.3 | C | 0.3 to C |
| C | 0.1 | A | 0.1 to A |
Attempt the task before opening its worked response. Record each change of state, explain the rule used, and check the result against the original requirements. Code examples labelled Python 3 use that language’s conventions; pseudocode states its own assumptions.
Common mistakes to avoid
Key takeaways
Test yourself
Answer a few questions, then review anything you missed.
Start with the quiz above to check the key ideas. You can also choose another topic from the practice library.