Information Retrieval

Informational Retrieval (IR) is the fetching and organising of info from a large number of text-based documents.

Measures of text retrieval #

{Retrieved} = set of docs retrieved

{Relevant} = set of docs relevant to the query

{Relevant \(\cap\) Retrieved} = set of docs retrieved and relevant

Precision #

$$ \text{precision} = \frac{\mid {Relevant} \cap {Retrieved}\mid}{\mid{Retrieved}\mid} $$

“Of the docs retrieved, how many were relevant to the query?”

Recall #

$$ \text{recall}=(|{Relevant}∩{Retrieved}|)/(|{Relevant}|)$$

“Of the relevant docs, how many were retrieved?”

F1 score #

$$ \text{F1}=\frac{2 \cdot (\text{recall} \times \text{precision})}{\text{recall} + \text{precision}} $$


March 4, 2025