Intelligent Semantic Web Search Engines: A Brief Survey

1. INTRODUCTION

• Keyword-based engines can struggle with intelligent queries, returning inaccurate results or results that may be unreliable.The introduction says these engines depend on information available in web pages.Source: §1 · p. 1
• The paper surveys existing intelligent semantic search methods and engines, classifying the literature and analyzing issues from different perspectives.The authors describe the survey as preliminary.Source: §1 · p. 2
• Semantic Web technologies are presented as a way to retrieve relevant and meaningful information more intelligently.The introduction says such technologies can bring results nearer to what users want.Source: §1 · p. 2

2. BACKGROUND

• The background identifies two challenges: mapping queries to meaningful information in documents and efficiently recognizing related results distributed across linked documents.It distinguishes query-to-document mapping from finding distributed results connected by hyperlinks.Source: §2 · p. 2
• The passage presents Semantic Web annotations, query-interface mechanisms, and ontologies as a way to produce intelligent and meaningful information, and describes knowledge representation and natural-language understanding as extraordinarily difficult problems.Source: §2 · p. 2
• The current Web lacks semantic structure, making it difficult for machines to understand user search strings and contributing to ambiguous or partly ambiguous results.The background also lists unstructured representation, poor interconnection, missing automatic transfer, trust, and machine-understanding concerns.Source: §2 · p. 3
• Hakia is described as a meaning-based search engine that seeks meaning matches rather than popularity of search terms.Its proprietary QDEXing technology processes presented news, blogs, credible sources, and galleries; Semantic Rank can process digital artifacts using third-party API feeds.Source: §2 · p. 3

3.1 Intelligent Search Engines

• Hung and Yang combine a description-logic inference system with a digital-library ontology, integrating knowledge inference and content search around user requirements.The account describes this combination as the basis of their intelligent search engine.Source: §3.1 · pp. 3–4
• Inamdar and Shinde describe personal agents that search on users’ behalf and recommend relevant web-page links to users and other agents.The agents can draw on different internal and external information sources and run on a server.Source: §3.1 · p. 4
• Lambrix, Shahmehri, and Wahllof use subsumption and default information in an extended description logic to retrieve documents on desired or more specific topics and typical topic content.Their system was tested on small-scale databases with promising results.Source: §3.1 · p. 4
• Prakash and colleagues describe simulation studies using a fuzzy satisfaction function and heuristic search criterion after modeling client behavior and web dynamics.Their architecture and design specifications address new-generation search engines and a knowledge framework for capturing intuition.Source: §3.1 · p. 4
• Meng and Huang describe an interactive model based on user information preferences, using artificial-intelligence methods to support individualized information search.The account presents the model as a way to improve information-retrieval quality and effectiveness.Source: §3.1 · p. 4
• A formal-concept-analysis search model combines information retrieval with concept-lattice browsing; tests validate feasibility and indicate the lattice can support flexible document management by conceptual relations.The model is based on formal context.Source: §3.1 · p. 4

4. TYPES OF SEMANTIC SEARCH ENGINES

• Semantic search engines are presented in the context of a Web whose information lacks structure and has ambiguity, limited automatic transfer, and challenges for machine understanding and trust.The section describes semantics as meaning that can be discovered by computers as well as people.Source: §4 · pp. 4–5

4.1 Semantic search engines

• XSEarch returns ranked, semantically related XML document fragments for simple queries, with experiments reporting efficiency, scalability, and highly ranked quality results.The account says its performance, recall, and precision were measured experimentally.Source: §4.1 · p. 5
• Eureka links files through an inference model and ranks them by semantic importance using a File Rank metric alongside keyword indices [20].Its crawler builds keyword and rank indices, which the query engine uses to match and rank files.Source: §4.1 · p. 5
• Wang et al. describe table search as identifying semantic relationships between cells, converting tables into database form, and retrieving data with query languages.Their method uses layout syntax grammar and templates to analyze cell semantics, followed by a semantic-preserving transformation.Source: §4.1 · pp. 5–6
• Avatar combines traditional text search with ontology annotations and transforms keyword queries into ranked interpretations whose retrieved documents are shown to users.Its components include a semantic optimizer and a user interaction engine.Source: §4.1 · p. 6

4.2 Ontology search engines

• Ontology-search examples include a registry for ontology metadata and a server for ontologies, with query-by-example and query-by-term search.SEWISE maps text from web sources into a uniform XML structure using a domain ontology, then applies text mining such as categorization and summarization.Source: §4.2 · p. 6

5. SOME COMMON ISSUES

• The authors describe the survey as preliminary and explicitly state that it is not comprehensive.Source: §5 · p. 6

a) Low precision and high recall

• Some semantic engines struggle to improve precision while lowering recall; Ding’s engine uses Google’s top 50 results, which could yield low precision and high recall.Source: a) Low precision and high recall · p. 6

b) Identity intention of the user

• Lee’s method analyzes request terms to fit user intention and make services more suitable, while individual user patterns can be extrapolated to global users.Source: b) Identity intention of the user · p. 6
• Users may omit synonyms or struggle to phrase a problem, while early disambiguation engines offered alternatives such as Java’s different meanings.Source: d) Inaccurate queries · p. 7, c) Individual user patterns can be… · p. 7

• The paper presents a brief survey of existing intelligent semantic search literature and reviews the methods’ characteristics.Source: §6 · p. 7
• The survey identifies four issue perspectives: designer–user perception differences, static knowledge structures, low precision with high recall, and lack of experimental tests.Source: §6 · p. 7
• Future work will pursue deeper and broader research to conclude the field’s current situation and promote further development of intelligent semantic search engines.Source: §6 · p. 7

Authors

• G. Madhu is pursuing a PhD in Computer Science and Engineering and works as a senior assistant professor in Information Technology at VNR VJIET.Source: Authors · p. 9
• A. Govardhan is Principal of JNTU Jagtial and has research interests including databases, data warehousing and mining, and information retrieval.Source: Authors · p. 9
• T. V. K. Rajinikanth is head of the Information Technology department at GRIET and researches data warehousing and mining, semantic web, spatial data mining, and ANN.Source: Authors · p. 9