IGNOU MLII-102 Solved Question Paper PDF Download

IGNOU MLII-102 Previous Year Solved Question Paper in Hindi

IGNOU MLII-102 Previous Year Solved Question Paper in English

Q1. What is a subject indexing language for information retrieval ? State its essential features with examples.

Ans. A Subject Indexing Language (SIL) is a system of controlled vocabulary and a set of rules used to represent the subject matter of information resources, such as books, articles, and other documents. Its primary purpose is to organize documents in an information retrieval system and help users access the information they need efficiently. It ensures consistency and accuracy in the indexing process, bridging the gap between the language of the author and the language of the user. An effective subject indexing language possesses the following essential features: 1. Vocabulary Control: This feature ensures that a single concept is represented by a single, authorized term. It resolves issues arising from synonyms (different words for the same concept), homonyms (same word with different meanings), and polysemous words. For instance, out of ‘Heart Attack’ and ‘Myocardial Infarction’, one is chosen as the standard term. Tools for vocabulary control include:

• Thesaurus: It shows relationships between terms, such as BT (Broader Term), NT (Narrower Term), and RT (Related Term). Example: MeSH (Medical Subject Headings).
• Subject Heading List: It is an alphabetical list of authorized terms. Example: LCSH (Library of Congress Subject Headings).

2. Syntax:

This refers to the grammar of the indexing language, which dictates how terms can be combined to form a meaningful subject statement. It clarifies the relationship between terms. For example, in PRECIS (Preserved Context Index System), terms are linked in a logical sequence using role operators to preserve the context of the subject.

3. Semantics:

This deals with the meaning of terms and the relationships between them. A good indexing language must clearly define what each term means and how it relates to other terms. The relationships used in a thesaurus, such as BT, NT, and RT, are examples of semantic control.

4. Hospitality:

The language must be flexible enough to accommodate new subjects and concepts. As knowledge is constantly evolving, an indexing language must have the capability to easily add new terms without disrupting the existing structure.

5. Literary Warrant:

The terms included in an indexing language should be based on the literature or documents of the subject they represent. This means that terms are included only when there is a real need for them, as evidenced by their appearance in documents.

Examples:

Sears List of Subject Headings (SLSH), Library of Congress Subject Headings (LCSH), and Medical Subject Headings (MeSH) are prominent examples of subject indexing languages used widely in libraries and information centers.

Or

Explain the concept of post-coordinate indexing. Describe uniterm indexing system. Explain post-coordinate search devices to avoid false drops.

Ans. Post-coordinate indexing is an indexing method where documents are analyzed and broken down into simple, single concepts or terms. These terms are called ‘keywords’ or ‘descriptors’. At the time of indexing, these terms are not combined but are recorded separately. The coordination or combination of terms is done by the user at the time of searching, which is why it is called ‘post-coordinate’. This is in contrast to pre-coordinate indexing, where the indexer combines terms to create complex subject headings beforehand. Uniterm Indexing System: This is a classic example of a post-coordinate indexing system, developed by Mortimer Taube in 1953. Its main features are:

• Analysis: The content of a document is analyzed and broken down into significant single terms, called ‘uniterms’.
• Uniterm Cards: A separate card is created for each uniterm.
• Posting: When a document is indexed with a uniterm, the document’s accession number is posted (recorded) on the corresponding uniterm card. The cards were often divided into 10 columns labeled 0-9, so that entries could be made based on the last digit of the document number. This technique, called ‘terminal digit posting’, speeds up the comparison of cards.
• Searching: To perform a search, the user selects the uniterm cards corresponding to their query. For example, to search for ‘Automation of Libraries in India’, the user would retrieve the cards for ‘Automation’, ‘Libraries’, and ‘India’. By comparing the document numbers on these cards, any number that appears on all the cards would represent a highly relevant document.

Post-coordinate Search Devices to Avoid False Drops:

False drops are irrelevant documents retrieved during a search because the search terms are present in the document but not in the relationship that the user intended. For example, a search for ‘Venetian blinds’ (a type of window covering) might retrieve a document about ‘blind Venetians’ (visually impaired people from Venice). To avoid this, the following devices are used:

1. Links:

A link is a code or symbol used to group related terms together within a single document. If a document discusses two distinct topics, all terms related to the first topic might be tagged with one link (e.g., Link A) and all terms for the second topic with another (e.g., Link B). During a search, only documents where all search terms are associated with the same link are retrieved, ensuring contextual relevance.

2. Roles:

Roles are indicators that specify the function or context of a term. For instance, the term ‘Iron’ could be used in the context of ‘export of iron’ or ‘import of iron’. A role indicator (e.g., ‘a’ for agent, ‘p’ for product) could clarify whether iron is the product being acted upon or the agent, thereby increasing the precision of the search.

3. Weighting:

This involves assigning a numerical value (weight) to index terms based on their importance in the document. At the time of searching, the system can rank documents higher if the combined weight of their terms matches the user’s query more closely. This helps prioritize the most relevant documents and reduces the impact of false drops.

---

Q2. Explain the features of the Colon Classification with special reference to its 7th edition (1987).

Ans. The Colon Classification (CC) , devised by Dr. S. R. Ranganathan, is a seminal work in library classification. It is the first major example of an analytico-synthetic and faceted classification scheme. The 7th edition, published posthumously in 1987, introduced several significant changes and refinements, solidifying its theoretical foundations. Its key features are: 1. Freely-Faceted Scheme: Unlike earlier editions that prescribed a rigid PMEST (Personality, Matter, Energy, Space, Time) formula, CC7 is described as a ‘freely-faceted’ scheme. The indexer first identifies the facets present in the subject of the document and then arranges them based on principles like the Wall-Picture Principle , Principle of a Whole Organ, etc. The PMEST formula serves as a helpful guide rather than a strict rule. 2. Introduction of Speciators: A major innovation in CC7 was the introduction of ‘Speciators’ to create compound isolates and subjects. Speciators are used to qualify a host isolate or facet.

• Speciator of Kind 1: Used to create a species/type of the host isolate. It is attached to the host isolate with a hyphen (-). Example: In ‘2-1-A’, ‘Library Science-Classification-Colon Classification’, ‘A’ (for CC) is a speciator of ‘1’ (Classification).
• Speciator of Kind 2: Used for properties or actions on the host isolate. It is attached with an ‘equals’ sign (=). Example: ‘D513=M’ could represent ‘Strength of a Beam’.

3. New Indicator Digits:

The 7th edition expanded the set of indicator digits to increase the scheme’s expressive power and hospitality. These include:

• Asterisk (*): Used to connect isolates in a subject string, indicating aggregation or a relationship.
• Ampersand (&): Used for Phase Relation, replacing the ‘0’ (zero) of previous editions.
• Quotation Mark (“): Used to introduce the common isolate for ‘System’ or ‘Special’.
• Apostrophe (‘): Used to introduce a Time Isolate.

4. Analytico-Synthetic Nature:

The core philosophy remains unchanged. The indexer analyzes the subject of a document into its constituent facets (analysis) and then combines these facets using the prescribed indicator digits to construct a specific class number (synthesis). This allows for the creation of co-extensive class numbers for highly specific subjects.

5. Designed for Computerization:

CC7 was designed with computer-based information retrieval in mind. Its logical structure, use of distinct indicators, and faceted nature make it theoretically more amenable to algorithmic manipulation than purely enumerative schemes.

6. Expansion of Main Classes and Isolates:

The 7th edition saw the revision and expansion of many schedules, including new main classes and a vast number of isolates to accommodate the growth of knowledge, particularly in science and technology.

Despite its theoretical elegance, the complexity of the 7th edition, its radical notational changes, and its incomplete publication (only Volume 1 was published) have limited its practical adoption. However, it remains a landmark in classification theory.

Or

Explain the features, structure and qualities of the UDC as an indexing language.

Ans. The Universal Decimal Classification (UDC) is a major international classification scheme developed by Paul Otlet and Henri La Fontaine at the end of the 19th century. It began as a French translation and expansion of the Dewey Decimal Classification (DDC) but evolved into a powerful and flexible indexing language. Its features, structure, and qualities make it highly suitable for modern information retrieval. Features:

• Universality: It aims to cover all fields of human knowledge, making it a universal classification system.
• Analytico-Synthetic Nature: While its main classes are enumerated like DDC, UDC is primarily analytico-synthetic. It allows for the combination of concepts from different parts of the schedules to create a precise class number for a compound subject.
• Internationality: It is managed by the UDC Consortium and is available in many languages, making it a truly international standard.
• Flexible Notation: UDC uses Arabic numerals (0-9) for its main classes, but its strength lies in a rich set of symbols (common auxiliary signs) that allow for detailed specification.

Structure:

The structure of UDC consists of two main parts:

1. Main Tables (Schedules):

These are the systematic tables of knowledge, divided into ten main classes (0-9), similar to DDC.

`0` – Generalities, Science and Knowledge

`1` – Philosophy, Psychology

`2` – Religion, Theology

`3` – Social Sciences

`4` – (Vacant)

`5` – Mathematics and Natural Sciences

`6` – Applied Sciences, Medicine, Technology

`7` – The Arts, Recreation, Entertainment, Sport

`8` – Language, Linguistics, Literature

`9` – Geography, Biography, History

2. Auxiliary Tables:

These are the key to UDC’s flexibility. They allow for the specification of recurring concepts like place, time, language, and form. There are two types:

• Common Auxiliaries: These can be used with any number from the main tables. They are identified by specific symbols, e.g., `=…` (Language), `(0…)` (Form), `(1/9)` (Place), `”…”` (Time), `(=…)` (Race and Nationality).
• Special Auxiliaries: These are specific to a particular main class or a section of it and are listed within the main tables. They are identified by notations like `-…`, `.0…`, and `’…`.

Example of Synthesis:

The subject “A monthly journal on railway engineering in French” would be synthesized as: `625.1` (Railway Engineering) `(051)` (Monthly Journal) `=112.2` (French Language).

Qualities as an Indexing Language:

• Flexibility and Expressiveness: The ability to combine numbers using auxiliaries makes UDC highly expressive and allows for the creation of very specific and co-extensive class numbers.
• Hospitality: The decimal notation and synthetic structure provide infinite hospitality, allowing new subjects to be inserted at any point.
• Suitability for Computerized Retrieval: The synthetic and logical structure of UDC class numbers makes them suitable for computerized systems. The defined syntax allows for precise searching, filtering, and browsing.
• Multi-disciplinary Indexing: The colon symbol `:` allows for the linking of two or more distinct UDC numbers to represent a complex, multi-disciplinary subject, a feature that is very powerful for retrieval.

---

Q3. What is UNICODE ? Describe its structure, features and the problems associated with it.

Ans. UNICODE is a universal character encoding standard that aims to represent every character used in modern and historical scripts in a single, unified character set. Before Unicode, there were hundreds of different encoding systems, which created conflicts and made it impossible to share data between different systems and languages reliably. Unicode was created to solve this “Mojibake” (garbled text) problem by providing a unique number for every character, no matter the platform, program, or language. Structure: The core of Unicode’s structure is the code point . A code point is a unique numerical value assigned to a single character or symbol. The Unicode standard defines a codespace of over one million (1,114,112 to be exact) possible code points, which are organized into 17 “planes”. Each plane contains 65,536 code points. To use these code points in computer systems, they must be encoded. The most common encoding forms are:

• UTF-8 (Unicode Transformation Format – 8-bit): This is a variable-width encoding that uses 1 to 4 bytes per character. It is backward compatible with ASCII (1-byte characters) and is the dominant encoding for the World Wide Web.
• UTF-16 (16-bit): A variable-width encoding that uses one or two 16-bit units. It is the native encoding for systems like Windows and Java.
• UTF-32 (32-bit): A fixed-width encoding that uses 4 bytes for every character. It is simple but uses more memory than UTF-8 and UTF-16.

Features:

• Universality: It covers almost all of the world’s written scripts, including modern languages, historical scripts, and technical symbols.
• Uniqueness: Each character is assigned a unique code point, ensuring there is no ambiguity.
• Uniformity: It provides a consistent way of representing and processing text, regardless of the underlying system. The encoding forms (UTF-8, etc.) provide well-defined methods for storing and transmitting Unicode data.
• Extensibility: The standard is continuously updated to include new scripts and characters as they are needed.

Problems Associated with Unicode:

• Complexity: Implementing Unicode correctly is complex. It involves more than just character mapping; it also requires handling complex rendering rules for scripts with ligatures (e.g., Arabic), combining diacritical marks (e.g., in Devanagari), and bidirectional text (e.g., Arabic and Hebrew).
• Storage Size: For text that is primarily in English or other Latin-based languages, UTF-8 is efficient. However, UTF-16 and UTF-32 can use significantly more storage space compared to older single-byte encodings like ASCII.
• Canonical Equivalence: Some characters can be represented in multiple ways. For example, ‘é’ can be a single pre-composed character or a combination of ‘e’ and a combining acute accent (´). This can cause problems in searching and comparison if not handled through a process called normalization.
• Legacy Compatibility: Converting legacy data from older encodings to Unicode can be challenging and can lead to data loss or corruption if not done carefully.

Or

What is Automatic Indexing ? Make a comparison between manual indexing and computerized indexing.

Ans. Automatic Indexing (also known as computerized indexing) is the process of using computer programs to analyze the text of a document and extract a set of index terms (keywords or phrases) that represent its subject content. The goal is to perform the indexing task with minimal or no human intervention, making it possible to process large volumes of documents quickly and consistently. There are several approaches to automatic indexing:

• Statistical Methods: These methods rely on the frequency and distribution of words in a document. The most common technique is TF-IDF (Term Frequency-Inverse Document Frequency) . It assumes that terms that appear frequently in a document but are rare in the overall collection are good descriptors of that document’s content. Other statistical methods include word co-occurrence analysis.
• Linguistic Methods: These methods use natural language processing (NLP) to analyze the grammatical structure and meaning of the text. This can involve part-of-speech (POS) tagging to identify nouns and noun phrases (which are often good index terms), syntactic parsing to understand relationships between words, and semantic analysis to handle synonyms and context.
• Hybrid Methods: Most modern systems use a combination of statistical and linguistic methods to achieve better accuracy.

Comparison between Manual Indexing and Computerized Indexing:

The choice between manual and computerized indexing depends on factors like the size of the collection, the required quality of indexing, budget, and time constraints. Often, a hybrid approach, where automatic indexing provides a first pass and human indexers refine the results, offers a good balance.

Factor	Manual Indexing	Computerized (Automatic) Indexing
Quality & Accuracy	Generally high. A human indexer understands context, nuance, metaphors, and can infer concepts not explicitly stated in the text. This leads to higher conceptual accuracy.	Variable quality. It is literal and based on the text alone. It cannot infer unstated concepts and may miss nuances. It often produces “noise” (irrelevant terms) and “silence” (missed relevant terms).
Consistency	Low. Consistency between different indexers (inter-indexer consistency) and even by the same indexer over time (intra-indexer consistency) can be a major problem.	High. The same algorithm applied to the same document will always produce the exact same set of index terms, ensuring perfect consistency.
Speed	Slow. It is a time-consuming intellectual process. A human can index only a limited number of documents per day.	Extremely fast. A computer can process thousands or millions of documents in a very short time.
Cost	High. It requires skilled, trained professionals, making it very expensive, especially for large collections.	Low operational cost. The initial investment in software and hardware can be high, but the per-document cost is very low, making it highly scalable.
Vocabulary	Can use either a controlled vocabulary (e.g., a thesaurus) for consistency or assign free-text keywords. The indexer can correctly map concepts to authorized terms.	Typically extracts terms directly from the document (natural language). While it can be programmed to map extracted terms to a controlled vocabulary, this can be complex and error-prone.

---

Q4. Explain the technical and economic factors in the evaluation of an information retrieval system.

Ans. The evaluation of an Information Retrieval (IR) system is crucial to determine its effectiveness and efficiency. This evaluation is typically based on a combination of technical and economic factors. Technical Factors Technical factors measure how well the system performs its core function of retrieving relevant information. The most important technical measures are: 1. Recall and Precision: These are the most famous evaluation metrics, often used together.

• Recall: The proportion of relevant documents in the collection that were successfully retrieved by the system. It measures the system’s ability to find all relevant items. Formula: Recall = (Number of relevant documents retrieved) / (Total number of relevant documents in the collection)
• Precision: The proportion of retrieved documents that are actually relevant. It measures the system’s ability to avoid retrieving irrelevant items (junk). Formula: Precision = (Number of relevant documents retrieved) / (Total number of documents retrieved)

There is often an inverse relationship between recall and precision; efforts to increase one may decrease the other.

2. Response Time:

This is the time taken by the system to respond to a user’s query. A slow response time can lead to user dissatisfaction, even if the results are good.

3. Throughput:

This measures the number of queries the system can handle simultaneously or per unit of time. It is a critical factor for systems with many concurrent users, such as web search engines.

4. User Effort:

This measures how easy it is for the user to interact with the system, formulate queries, and understand the results. It includes factors like the quality of the user interface (UI) and the complexity of the query language.

5. Coverage:

The extent to which the system’s database includes the documents and sources that are relevant to its target user population.

6. Fallout:

The proportion of non-relevant documents retrieved in relation to all non-relevant documents in the collection. It is less commonly used than precision but provides another measure of retrieval of non-relevant items.

Economic Factors

Economic factors relate to the costs and benefits of implementing and operating the IR system. They are crucial for justifying the system’s existence and continued funding.

1. Cost-Benefit Analysis:

This is the overarching economic evaluation. It compares the total costs of the system with its total benefits. Benefits can be tangible (e.g., saved staff time, reduced subscription costs) or intangible (e.g., better decision-making, increased user satisfaction). The system is economically viable if the benefits outweigh the costs.

2. Hardware and Software Costs:

This includes the initial procurement cost of servers, storage, networking equipment, and the IR software itself (licensing or development costs).

3. Operational Costs:

These are recurring costs required to keep the system running. They include:

• Staff Costs: Salaries for system administrators, indexers, and support staff.
• Maintenance Costs: Costs for hardware and software maintenance contracts, updates, and repairs.
• Data Acquisition Costs: Costs for acquiring or subscribing to the content (databases, journals) that populates the system.
• Energy and Facility Costs: Power consumption for servers and data centers.

4. User Training Costs:

The cost associated with training users to effectively use the system. A complex system may incur higher training costs.

5. Cost-Effectiveness:

This compares the costs of different systems that achieve the same level of performance (e.g., the same recall and precision). The goal is to choose the system that achieves the desired technical performance at the lowest possible cost.

Or

Enumerate any six information retrieval models based on theories and tools and explain any three of them in detail.

Ans. An information retrieval (IR) model provides a framework for defining the structure of documents and queries, and a ranking mechanism to determine the relevance of a document to a query. Different models use different theories and tools to achieve this. Six prominent information retrieval models are: 1. Boolean Model 2. Vector Space Model (VSM) 3. Probabilistic Model 4. Language Model 5. Latent Semantic Indexing (LSI) Model 6. Fuzzy Set Model Here are three of these models explained in detail: 1. Boolean Model The Boolean model is the earliest and simplest IR model. It is based on set theory and Boolean algebra.

• Theory: Documents and queries are treated as sets of terms. The model uses Boolean operators— AND , OR , and NOT —to formulate queries.
• Process:
  • A query like “information AND retrieval” will only retrieve documents that contain both terms.
  • “information OR retrieval” will retrieve documents containing either term (or both).
  • “information NOT science” will retrieve documents containing “information” but not “science”.
• Tools & Strengths: Its main strength is its precision and predictability. It is well-understood by expert users like librarians and lawyers who need to construct precise queries.
• Weaknesses: The model is very rigid. It does not provide any mechanism for ranking documents; a document is either a match or not. It is difficult for novice users to formulate effective Boolean queries, and it does not handle partial matches. A query with too many ANDs may yield no results, while one with too many ORs may yield too many.

2. Vector Space Model (VSM)

The Vector Space Model (VSM) overcomes the major limitations of the Boolean model by introducing the concept of partial matching and ranked results.

• Theory: It represents documents and queries as vectors in a multi-dimensional space, where each dimension corresponds to a unique term in the collection.
• Process:
  1. Term Weighting: Each term in a document vector is assigned a weight that indicates its importance in that document and in the collection as a whole. The most common weighting scheme is TF-IDF (Term Frequency-Inverse Document Frequency) . TF measures how often a term appears in a document, while IDF measures how rare the term is across the entire collection.
  2. Vector Representation: A document D is represented as a vector `V(D) = (w1, w2, …, wn)`, where `wi` is the TF-IDF weight of the i-th term. The query is also converted into a similar vector.
  3. Similarity Calculation: The relevance of a document to a query is determined by calculating the similarity between their vectors. The most common method is cosine similarity , which measures the angle between the two vectors. A smaller angle (cosine value closer to 1) means higher similarity.
• Tools & Strengths: VSM’s main strength is its ability to rank documents according to their similarity to the query, providing users with a prioritized list of results. It handles partial matches effectively.
• Weaknesses: It assumes that terms are independent, which is not always true (e.g., “information” and “retrieval” are not independent). It is also computationally more intensive than the Boolean model.

3. Probabilistic Model

The Probabilistic Model is based on the idea of ranking documents in order of their probability of being relevant to a user’s query.

• Theory: The model is built upon the Probability Ranking Principle (PRP) , which states that an IR system will achieve optimal performance if it ranks documents in decreasing order of their probability of relevance to the user’s query.
• Process:
  1. The system initially guesses a set of relevant documents for a query.
  2. It then analyzes the term distribution in this initial set (assumed relevant) and the rest of the collection (assumed non-relevant).
  3. It calculates the probability that a document `D` is relevant to a query `Q`, i.e., `P(R|D)`. Using Bayes’ Theorem, this is used to create a ranking function.
  4. The system uses these probabilities to rank all documents. The process can be iterated and improved using relevance feedback , where the user identifies some retrieved documents as relevant, allowing the system to refine its probabilistic estimates and provide a better-ranked list.
• Tools & Strengths: Its theoretical foundation in the PRP is sound. It provides a ranked list of results and is one of the most effective models, especially when relevance feedback is incorporated. The BM25 (Best Match 25) ranking function, a refinement of this model, is widely used in modern search engines.
• Weaknesses: The initial assumption of a relevant set is a key challenge. It also assumes term independence, similar to VSM. The underlying mathematical theory can be complex to understand and implement.

---

Q5. Write short notes on any three of the following in about 300 words each : (a) Expert Systems (b) Uses of Citation Indexing (c) ISBDs (d) Functions of Thesaurus (e) MEDLARS test

Ans. (a) Expert Systems An Expert System is a computer program from the field of Artificial Intelligence (AI) designed to emulate the decision-making and problem-solving capabilities of a human expert in a specific, narrow domain of knowledge. Instead of being programmed with procedural logic, it is programmed with facts and rules about a particular subject. The core components of an expert system are:

• Knowledge Base: This is the heart of the system. It contains the facts, data, and rules-of-thumb (heuristics) that a human expert would use. This knowledge is often represented in the form of “IF-THEN” rules (e.g., “IF the patron is an undergraduate AND the book is on reserve, THEN the loan period is 2 hours”).
• Inference Engine: This is the “brain” of the system. It is a generic reasoning mechanism that applies the rules in the knowledge base to the facts of the current problem to arrive at a conclusion or recommendation. It uses techniques like forward chaining (data-driven) or backward chaining (goal-driven).
• User Interface: This allows a non-expert user to interact with the system, enter details about a problem, and receive the expert advice or solution.

In the context of libraries and information science, expert systems have been developed for various applications, such as:

• Reference Services: A reference expert system can guide a user through a series of questions to help them identify relevant databases or reference books for their query (e.g., POINTER).
• Cataloging: An expert system can assist a cataloger by suggesting subject headings or correctly formatting a bibliographic record based on cataloging rules like AACR2 or RDA.
• Indexing and Classification: They can help in assigning appropriate index terms or classification numbers by interpreting the subject content of a document based on a set of rules.

While the initial hype around expert systems in the 1980s has subsided, the principles behind them continue to influence modern knowledge-based systems, recommendation engines, and AI applications.

(b) Uses of Citation Indexing

Citation Indexing

is a unique form of indexing that connects documents based on the citations between them. Instead of using keywords or subjects, it uses the bibliography (or reference list) of one document to link it to other documents. Conceived by

Eugene Garfield

, it led to the creation of the Science Citation Index (SCI) and is now the foundation of major databases like Web of Science and Scopus.

The premise is that when one author cites another’s work, a subject relationship is established between the two documents. This creates a powerful network of scholarly communication with numerous uses:

• Literature Search and Discovery: This is its most fundamental use. Starting with a single relevant paper, a researcher can move backward in time by looking at its references ( backward chaining ) and forward in time to see who has cited that paper since its publication ( forward chaining ). This allows for a comprehensive literature review and the discovery of related, more recent research.
• Measuring Research Impact: The number of times a paper or an author is cited is widely used as a proxy for its influence and impact in a field. This is the basis for metrics like the Journal Impact Factor and the author h-index.
• Identifying Seminal Works: Highly cited papers within a discipline are often considered foundational or seminal works. Citation analysis can quickly identify these key publications.
• Interdisciplinary Research Discovery: Citation indexing can reveal unexpected connections between different fields of study when a paper in one discipline is cited by a paper in a completely different discipline.
• Author Evaluation and Tenure/Promotion: In academia, citation counts and derived metrics are often used by institutions to evaluate the research output of faculty for hiring, promotion, and tenure decisions.
• Bibliometric Studies: It provides the raw data for bibliometrics, the statistical analysis of publications, which can be used to map the structure of science, identify research trends, and analyze collaborative patterns.

Despite its power, citation indexing has limitations, including potential bias towards older papers and different citation practices across disciplines.

(c) ISBDs (International Standard Bibliographic Description)

The

International Standard Bibliographic Description (ISBD)

is a set of rules developed by the International Federation of Library Associations and Institutions (IFLA) to create bibliographic descriptions in a standardized, human-readable format. Its primary goal is to aid the international exchange of bibliographic records by making them identifiable and understandable regardless of the language of the description or the script it is written in.

The key features of ISBD are:

• Defined Areas: The ISBD specifies a standard order for the elements of a bibliographic description, grouping them into distinct areas. The main areas are:<
  • Area 0: Content form and media type area
  • Area 1: Title and statement of responsibility area
  • Area 2: Edition area
  • Area 3: Material or type of resource specific area
  • Area 4: Publication, production, distribution, etc., area
  • Area 5: Physical description area
  • Area 6: Series area
  • Area 7: Notes area
  • Area 8: Resource identifier and terms of availability area
• Prescribed Punctuation: The most recognizable feature of ISBD is its use of prescribed punctuation. Each area is preceded by a specific punctuation mark or symbol (e.g., a period and a space `. `). Within areas, different elements are separated by specific symbols (e.g., `/` for statement of responsibility, `:` for other title information, `;` to separate multiple statements). This punctuation acts as a code, allowing a person or a computer to interpret the record even without understanding the language.

For example: `Title : other title information / first statement of responsibility ; second statement. –- Edition statement. –- Place of publication : Publisher, date.`

The ISBD is not a cataloging code itself; rather, it provides the framework for the descriptive part of cataloging. Major cataloging codes like AACR2 (Anglo-American Cataloguing Rules, 2nd edition) and RDA (Resource Description and Access) are based on the principles and structure of ISBD. Its implementation was a major step towards Universal Bibliographic Control (UBC).

(d) Functions of a Thesaurus

A

thesaurus

in the context of information retrieval is a controlled vocabulary tool that organizes terms and shows the semantic relationships between them. It is a crucial instrument for both indexers and searchers, designed to improve the effectiveness of an information system by bridging the gap between the terminology of documents and the terminology of users.

A thesaurus performs several key functions:

1. Vocabulary Control:

This is its primary function. It controls synonyms and quasi-synonyms by establishing a single preferred term (or ‘descriptor’) to represent a concept. Other synonyms are listed as non-preferred terms and point to the descriptor.

• Example: `Myocardial Infarction` USE `Heart Attack`. This ensures that all documents about the concept are indexed under the same term.

2. Clarifying Relationships:

It explicitly defines the relationships between concepts, which helps in both indexing and searching.

• Hierarchical Relationship (BT/NT): It shows broader (BT) and narrower (NT) terms, creating a conceptual hierarchy. E.g., `Vehicles` (BT) -> `Cars` (NT). This allows a searcher to broaden or narrow their search.
• Associative Relationship (RT): It links related terms (RT) that are not hierarchically related but are conceptually close. E.g., `Birds` RT `Ornithology`. This suggests alternative search paths.
• Equivalence Relationship (USE/UF): It handles synonyms by directing the user from a non-preferred term (UF – Used For) to the preferred descriptor (USE).

3. Aiding Indexers:

A thesaurus provides a standardized list of terms for indexers to choose from. This ensures inter-indexer consistency, meaning different indexers will use the same term for the same concept, which is vital for the quality of the database.

4. Assisting Searchers:

It serves as a user’s guide to the database. By browsing the thesaurus, a searcher can understand the vocabulary used in the system, find the most appropriate terms for their query, and discover broader, narrower, or related terms they may not have considered, thus improving search strategies.

By performing these functions, a thesaurus helps to improve both

recall

(by linking related terms and synonyms) and

precision

(by disambiguating terms and providing specific descriptors).

(e) MEDLARS test

The

MEDLARS test

was a landmark large-scale evaluation of an operational information retrieval system, conducted by F.W. Lancaster in the mid-1960s. The system under evaluation was MEDLARS (Medical Literature Analysis and Retrieval System), a pioneering computerized bibliographic database created by the U.S. National Library of Medicine (NLM). This test is significant because, unlike earlier experiments like Cranfield which were conducted in a controlled laboratory setting, the MEDLARS test evaluated a real-world, large-scale system in its operational environment, using real user queries.

Key aspects of the test:

• Objective: The primary goal was to evaluate the performance of the MEDLARS system in terms of its ability to satisfy the information needs of its users.
• Methodology: Lancaster collected 300 real search requests that had been processed by the system. For each request, he, with the help of the original requesters, conducted an exhaustive manual search of the literature to identify all relevant documents in the database, thus establishing a ‘gold standard’ for relevance. He then compared the system’s output against this standard to measure its performance.
• Evaluation Metrics: The test measured the system’s performance using the standard IR metrics of Recall (the proportion of relevant documents retrieved) and Precision (the proportion of retrieved documents that were relevant).
• Failure Analysis: A crucial part of the study was the detailed analysis of retrieval failures. Lancaster didn’t just calculate recall and precision scores; he investigated why the system failed to retrieve relevant items (recall failures) or why it retrieved irrelevant ones (precision failures).

Findings and Significance:

The overall performance was found to be an average of 58% recall and 50% precision. More importantly, the failure analysis revealed that the majority of retrieval failures were due to human factors, not machine issues. The main causes were identified as:

• Indexing: Inconsistent or incomplete indexing.
• Search Formulation: Ineffective translation of the user’s information need into a search query.
• User-System Interaction: Lack of sufficient interaction between the searcher and the end-user.
• Vocabulary Issues: Deficiencies in the system’s controlled vocabulary (MeSH).

The MEDLARS test was a pioneering effort in the evaluation of operational IR systems and its methodology and findings had a profound impact on the field, highlighting the critical importance of the human element (indexing and search formulation) in system performance.