<p>Shiv Nadar University is a comprehensive, multidisciplinary, research-focused, and student-centric University offering a full range of academic programs at the undergraduate, postgraduate and doctoral level. With state-of-the-art infrastructure, the University comprises of academic wings, sophisticated labs, international standard sports facilities, amphitheaters, auditorium, conference rooms and smart classrooms. The University&rsquo;s goal is to become internationally recognized for the quality of its research and creative endeavors and their applicability to improving quality of life, generating new insights and expanding the boundaries of human knowledge creativity. Committed to excellence in teaching, research and service, the University aims to serve the higher education needs of India and the world beyond.</p>

Shiv Nadar University

In this work we have extended and implemented an ontology based approach for email classification based on user characteristics proposed by Kim et al.(2007). The approach focuses on finding relationships between user interests and their responses to emails. Rules and Ontology are created using the data and metadata of user characteristics, their preferences and responses to emails. Rules and ontology are then used to predict the response of a user to a new email. In Kim et al. (2007) approach, labels to emails were provided manually by a human expert. We have endeavored to remove the human intervention by developing an Automated Email Categorizer to provide label to an email based on its contents. We have also proposed a new term weighing method for emails to incorporate prominence of subject terms. Finally, we have integrated and tested the Ontology Based Classifier in conjunction with Email Categorizer where the former effectively uses the label provided by latter to classify an email based on user preferences.

World University root node

Computer Science and Engineering

(SoE) School of Engineering

KEOD 2010 - Proceedings of the International Conference on Knowledge Engineering and Ontology Development

Automatic email classification using user preference ontology

This paper designs a novel fuzzy attributes and competition based bidding strategy (FAC-Bid), in which the final best bid is calculated on the basis of the assessment of multiple attributes of the goods and the competition for the goods in the market. The assessment of attributes adapts the fuzzy sets technique to handle uncertainty of the bidding process. The bidding strategy also uses and determines competition in the market (based on the two factors i.e. no. of the bidders participating and the total time elapsed for an auction) using Mamdani's Direct Method. Then the final price of the best bid will be determined based on the assessed attributes and the competition in the market using fuzzy reasoning technique.

ICEIS 2010 - Proceedings of the 12th International Conference on Enterprise Information Systems

Using fuzzy set approach in multi-attribute automated auctions

A digital repository is a digital archive for depositing and providing access to digital contents. It harvests metadata for providing an access framework of digital content to multiple users. In existing systems, query and retrieval is done on centralized basis and do not consider semantics of the query. Since the repositories are distributed, distinct and autonomous, there is a need for a framework which is capable of providing an efficient integrated and semantically enriched query retrieval mechanism over multiple repositories. We have proposed and implemented a multi-agent based framework for querying multiple disjoint repositories that is equipped with well defined semantic web technologies such as ontology [7, 8]. The system has been implemented using Jena framework for semantic web, Jade for agents and Java Swing for user interfaces. © 2010 IEEE.

2010 International Conference on Machine and Web Intelligence, ICMWI 2010 - Proceedings

Multi-agent based architecture for querying disjoint data repositories

Web services provide a promising solution to an age old need of fast and flexible information sharing among people and businesses. One of the key research issues in web services is selection of most pertinent web services. Selection of web service has become a tedious job because of the increasing number of service providers providing services with similar functionality. Therefore, a framework for selection and discovery of web service that can meet the non-functional requirements is needed. In this work, we have developed a selection tool (WSS-NFP) for web services which can rank services based on their non functional properties such as performance, delay etc. We have used soft computing techniques for selection and discovery of web services based on non functional properties. These properties are fuzzy in nature. A tool is developed and implemented that is generic in nature and can be customized for any application domain. As a core of this tool, it consists of Fuzzy System for ranking web services and Neural Network for fine tuning the membership functions for linguistic terms. Appropriate and fine tuned membership functions help in minimizing the output error measure and maximizing the performance index of the proposed tool. This tool is basically a Neuro-Fuzzy system for ranking web services of any domain. © 2010 IEEE.

WSS-NFP: Tool for web service selection based on non-functional properties using soft computing

In this paper, we have presented a new approach to predict pattern of the financial time series in stock market for next 10 days and compared it with the existing method of exact value prediction [2, 3, and 4]. The proposed pattern prediction technique performs better than value prediction. It has been shown that the average for pattern prediction is 58.7% while that for value prediction is 51.3%. Similarly, maximum for pattern and value prediction are 100% and 88.9% respectively. It is of more practical significance if one can predict an approximate pattern that can be expected in the financial time series in the near future rather than the exact value. This way one can know the periods when the stock will be at a high or at a low and use the information to buy or sell accordingly. We have used Support Vector Machine based prediction system as a basis for predicting pattern. MATLAB has been used for implementation. © Springer-Verlag Berlin Heidelberg 2009.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Pattern prediction in stock market

The simplicity and proliferation of World Wide Web (WWW) has taken the availability of information to an unprecedented level. Moreover, this entire available information is in unstructured textual format that results search process inefficient. Simultaneously, development of web technology led to the new concept of semantic web [1], which is an extension of current web and seeks to make information understandable by machines. This creation of semantic web content involves semantic annotation of the web pages, extraction of essential data from the annotated web page and finally generation of semantic web page through processes that require high degree of engineering skill. We have developed integrated tool that helps in creation of semantic web page based on ontology [4] from HTML web page, thus enabling computers and people to work in cooperation in a better way. Semantic web page will help to process user's query in meaningful way. Copyright © 2007 IICAI.

Proceedings of the 3rd Indian International Conference on Artificial Intelligence, IICAI 2007

Automatic conversion of HTML pages to semantic web pages

A Question Answering (QA) system aims to return exact answers to natural language questions. While today information retrieval techniques are quite successful at locating within large collections of documents those that are relevant to a user's query, QA techniques that extract the exact answer from these retrieved documents still do not obtain very good accuracies. We approached the TREC 2007 Question Answering task as a semantics based question to answer matching problem. Given a question we aimed to extract the relevant semantic entities in it so that we can pin-point the answer. In this paper we show that our technique obtains reasonable accuracy compared to other systems.

NIST Special Publication

IITD-IBMIRL system for question answering using pattern matching, semantic type and semantic category recognition

Contact center agents typically respond to email queries from customers by selecting predefined answer templates that relate to the questions present in the customer query. In this paper we present a technique to automatically select the answer templates corresponding to a customer query email. Given a set of query-response email pairs we find the associations between the actual questions and answers within them and use this information to map future questions to their answer templates. We evaluate the system on a small subset of the publicly available Pine-Info discussion list email archive and also on actual contact center data comprising customer queries, agent responses and templates.

IJCAI International Joint Conference on Artificial Intelligence

Automatically selecting answer templates to respond to customer emails

Automatic filtering of information has become increasingly important in recent years due to large availability of electronic information. We present an approach for optimization of Information Retrieval System (IRS) by genetic algorithm and fuzzy sets in an adaptive filtering process. From the observed user preferences regarding documents in the sets retrieved, the system learns from the user's information needs. An interest profile is created that represents the needs as learned from the observed preferences in the user's area of interest. The proposed system will act as an offline information-filtering agent. The documents have been already downloaded from the Internet using google. The system generates a recommendation based on adaptive filtering using a set of keywords extracted from all documents evaluated by user. The process starts with the initial set of documents retrieved as the answer to user's initial query in the area of interest. The preferences given by the user are learned through explicit feedback on retrieved documents. The agent filters and ranks the retrieved information according to user's preferences using Genetic Algorithm (GA) and Fuzzy Set Theory. The system has been implemented using Java on Windows 98. © 2006 Taylor & Francis Group, LLC.

IETE Journal of Research

Information filtering using fuzzy-genetic algorithm approach

The world of text is increasing considerably. Hence, a successful summarization technology has a key role to play in the future of computing. This paper presents an efficient method of intelligently generating summaries, which uses operations derived from an analysis of human written abstracts. The algorithm uses both statistical and semantic based heuristics to produce cohesive extracts. In the proposed system, an efficient method for evaluating lexical chains as an intermediate step has been used. We have also introduced a method for finding text cohesion and implemented the same in the summarizer. The proposed summarizer also edits long extracted sentences, using reduction to remove non essential phrases. The system has been implemented using JAVA standard Development kit. Copyright © IICAI 2005.

Proceedings of the 2nd Indian International Conference on Artificial Intelligence, IICAI 2005

Automatic text summarization

A hybrid information extraction (IE) framework based on geographical term detection approach has been developed to extract geographical information from an unrestricted Hindi text The relationship between geographical entities extracted with the adjacent text is shown graphically so that information about these entities can be related. The system, a combination of statistically and linguistically motivated techniques, identifies single geographical names and multiple geographical names as well. The method is applied on Hindi language text, but the approach can be adapted for other languages also. The paper presents some experiments illustrating the accuracy of the method. The system being developed is in a prototype stage and will be extended to include relation mark-up as well. © 2005 IEEE.

Proceedings of 2005 IEEE International Conference on Natural Language Processing and Knowledge Engineering, IEEE NLP-KE'05

Hybrid framework for information extraction for geographical terms in hindi language texts

The advent of e-commerce has brought about a radical change in the process of auctions. This has given bidders flexibility to try and get best possible deals by participating in multiple auctions. This has also provided sellers with the opportunity of unlawfully bidding (skill bidding) in their auctions with the sole purpose of increasing their payoffs. This paper looks at the concept of shill bidding and also introduces another concept (mole bidding) through which seller can increase their payoffs. The proposed solution is based on software agents. This paper shows a high level decision making framework for sellers using mole agent (automated alias for a mole bidder) to bid across various overlapping English auctions with varying start and end times. © 2005 IEEE.

Proceedings - International Conference on Computational Intelligence for Modelling, Control and Automation, CIMCA 2005 and International Conference on Intelligent Agents, Web Technologies and Internet

Mole agents - A new phenomenon in auctions (from sellers point of view)

The paper presents an information extraction system that takes input from Hindi texts and improves the information content retrieved by using anaphor/pronoun resolution mechanism. The information extraction system developed consists of three major modules: The language Parser, Resolution System and Information Extractor. The language parser used is HPSG (Head-Driven Phrase Structure Grammar) based that provides both syntactic and semantic information to the anaphor resolution system. HPSG was chosen because it provides a set of constraint on the co-referential structures in the language, which bounds the search for an antecedent to a more precise location in the discourse. The semantic information included in its parsing may be helpful for removing ambiguity in anaphor/pronoun resolution. The anaphor resolution system uses few heuristic rules to resolve intrasentential references while centering theory is used for intersentential resolution.

Proceedings of the 4th International Conference on Language Resources and Evaluation, LREC 2004

Information extraction from Hindi texts

The advent of e-commerce has brought about a radical change in the process of auctions. There is a trend towards automation in the process of auctions. Agents are widely used in this automation process. Case-based reasoning is a process which provides a user a method to automate the process of bidding. This process has an inherent learning component. CBR combined with a bidding agent for strategizing the bidding functionality provides a flexible solution for automating the bidding process for any type of auction in any domain. This paper proposes a generic bidding agent which can participate in the auction. The proposed solution is based on CBR learning technique and software agents.

Proceedings of the IASTED International Conference. Applied Informatics

CBR based automatic bidding in auctions

In this paper a study on the references is performed. An algorithm has been developed to find the referent of implicit anaphora. A representation of noun storage in knowledgebase is devised. The algorithm is designed to identify the presence of implicit anaphora, and is implemented in C++. A prototype lexicon has been designed. The algorithm worked well for test inputs. The results are presented. © 2003 Taylor & Francis Group, LLC.

Pronominal referent interpretation system for english text

An algorithm, implementable in parallel, for sign detection in nonredundant residue number systems is presented. The new approach in this paper completely avoids the expensive process of the mixed radix conversion (MRC), yielding a significant saving in the process time. An advancement in LSI technology supports the possible parallel realization of the algorithm. © 1986.

Computers and Electrical Engineering

Sign detection in residue code

A new approach for correcting multiple errors and detecting an additive overflow in the Residue Number System (RNS) is suggested. It works with the code whose redundancy is in the form of magnitude indices. The residue representation of a number with magnitude index is reconsidered. The RNS with magnitude index were first studied by Sasaki16 and Rao15 and then by Barsi and Maestrini5, 6. The range of a given RNS is divided into intervals of equal width and the magnitude of a number X is defined as a integer locating X into one of such intervals. We have proposed algorithm which detects and corrects multiple errors in residue number. The algorithms for special cases viz., single burst residue error and single residue error are also suggested. Some of the advantages are pointed out over the existing approaches. © 1985 IEEE.

Proceedings - Symposium on Computer Arithmetic

Multiple error correction and additive overflow detection with magnitude indices in residue code

This paper describes the general schemes for implementing the basic modular operations of addition, subtraction and multiplication in the Residue Number System (RNS). We use semiconductor logic gates in the proposed implementations. A comparative study has been carried out between the proposed schemes and the schemes known hitherto. © 1984.

Implementation schemes for basic modular operations

SIGN DETECTION IN NON-REDUNDANT RESIDUE NUMBER SYSTEM WITH REDUCED INFORMATION.

In this paper, an implementation scheme involving decoders and residue adders has been suggested to convert input data in fixed radix representation to residue representation. A method dealing with the reverse process is also demonstrated. A comparison has been made with the methods known hitherto. © 1982.