Search Engines - Recall versus Relevance

Saturday, March 18th 2023

In the context of developing search engines, recall and relevance are two important metrics that are often used to evaluate the effectiveness of search algorithms.

Recall refers to the percentage of relevant documents that are retrieved by the search engine out of the total number of relevant documents in the collection. In other words, recall is a measure of how well the search engine is able to retrieve all relevant documents in response to a user's query.

Relevance, on the other hand, refers to how well the retrieved documents match the user's query. Relevance is often measured by the degree to which the retrieved documents satisfy the information need expressed in the user's query.

In general, a good search engine should aim for high levels of both recall and relevance. However, achieving high levels of both can be challenging as there is often a trade-off between the two metrics. For example, improving recall may lead to a larger number of irrelevant documents being retrieved, while improving relevance may lead to some relevant documents being missed.

Other metrics

In addition to recall and relevance, there are other metrics to consider when evaluating the effectiveness of a search engine. Some of these include:

  1. Precision: This measures the proportion of relevant results among the retrieved results. A high precision means that the majority of the retrieved results are relevant.

  2. F1 score: This is a weighted average of precision and recall, which provides a balance between the two measures.

  3. Click-through rate (CTR): This measures the percentage of users who clicked on one of the search results. A high CTR suggests that the results are relevant and useful to users.

  4. Dwell time: This measures the amount of time users spend on a page after clicking on a search result. A high dwell time indicates that the page is relevant and engaging to users.

  5. Query latency: This measures the time it takes for the search engine to respond to a query. A fast response time is important for providing a good user experience.

  6. Index freshness: This measures how frequently the search engine updates its index to include new content. A fresh index is important for ensuring that the search results are up-to-date.

Overall, a good search engine should strive to optimize all of these metrics to provide the best possible user experience.


In the context of search engines, precision is a measure of how many of the retrieved documents are relevant to the user's search query. It is defined as the number of true positive results divided by the number of true positives plus the number of false positives. In other words, precision measures the proportion of relevant documents among the documents retrieved by the search engine.

For example, if a search engine retrieves 100 documents and 70 of them are relevant, while the remaining 30 are irrelevant, the precision would be 70/(70+30) = 0.7, or 70%.

Precision is an important metric for evaluating search engines because it directly reflects the usefulness of the results to the user. A high precision indicates that the search engine is effectively retrieving relevant documents, while a low precision indicates that many of the retrieved documents are irrelevant to the user's query.

F1 Score

In the context of search, F1 score is a metric used to evaluate the effectiveness of a search engine's results. It is a measure of the trade-off between precision and recall. Precision refers to the proportion of relevant results among all the results returned by the search engine, while recall refers to the proportion of relevant results among all the relevant results in the dataset.

The F1 score is the harmonic mean of precision and recall, and provides a way to evaluate a search engine's performance on a single metric that balances both precision and recall. It is calculated as follows:

F1 score = 2 * (precision * recall) / (precision + recall)

An F1 score of 1.0 indicates perfect precision and recall, while a score of 0.0 indicates that no relevant results were returned by the search engine. In general, higher F1 scores indicate better search engine performance.