JNTU MCA 2021 INFORMATION RETRIEVAL SYSTEMS Previous Question Paper (February - 2021 - R17)

Information retrieval (IR) and data retrieval (DR) differ fundamentally in their objectives, precision, and handling of uncertainty. Data retrieval focuses on finding specific records in a database that match a precise query, where the result is typically binary—either a record matches the criteria or it does not. In contrast, information retrieval deals with unstructured or semi-structured data, such as natural language text, where the goal is to find documents that are relevant to a user's information need.

Key differences include:

• Matching: DR uses exact matching based on Boolean logic or relational algebra, while IR uses partial or fuzzy matching based on similarity measures or statistical models.
• Precision: DR provides high precision with no ambiguity; IR involves uncertainty, as relevance is subjective and depends on the user's intent.
• Query Language: DR uses formal, structured query languages like SQL, whereas IR uses keyword-based queries or natural language.
• Evaluation: DR is evaluated by the correctness of the retrieved set, while IR is evaluated using metrics like precision, recall, and F-measure.

The logical view of a document is an abstraction that represents the document in a form suitable for the retrieval system. It transforms raw text into a structured format that the system can process efficiently to determine relevance. This process typically involves several stages of text processing:

• Tokenization: The document is broken down into individual units called tokens, such as words or phrases, by removing punctuation and handling whitespaces.
• Stop-word Removal: Common words that carry little semantic value (e.g., 'the', 'is', 'at') are removed to reduce index size and noise.
• Stemming or Lemmatization: Words are reduced to their root form (e.g., 'running' and 'runs' become 'run') to ensure that variations of the same word are treated as the same concept.
• Weighting: Each term is assigned a weight, often using techniques like TF-IDF (Term Frequency-Inverse Document Frequency), which reflects the importance of a term within a specific document relative to the entire collection.

The resulting logical view is often stored as an inverted index, which maps terms to the list of documents containing them, enabling rapid searching during the retrieval process.