R. Subhashini

Evaluating the Performance of Similarity Measures Used in Document Clustering and Information Retrieval

This paper presents the results of an experimental study of some similarity measures used for both Information Retrieval and Document Clustering. Our results indicate that the cosine similarity measure is superior than the other measures such as Jaccard measure, Euclidean measure that we tested. Cosine Similarity measure is particularly better for text documents. Previously these measures are compared with the conventional text datasets but the proposed system collects the datasets with the help of API and it returns the collection of XML pages. These XML pages are parsed and filtered to get the web document datasets. In this paper, we compare and analyze the effectiveness of these measures for these web document datasets.

Shallow NLP techniques for noun phrase extraction

The field of Information Retrieval plays an important role in searching on the Internet. Most of the information retrieval systems are limited to the query processing based on keywords. In information retrieval system the matching of the query against a set of text record is the core of the system. Retrieval of the relevant natural language text document is of more challenge. Todays most search engines are based on keyword based (bag of words) techniques, which results in some disadvantages. For text retrieval key phrases can help to narrow the search results or rank retrieved documents. We exploit shallow NLP techniques to support a range of NL queries and snippets over an existing keyword-based search. This paper describes a simple system for choosing noun phrases from a document as key phrases. The noun phrase extractor is made up of three modules: tokenization; part-of-speech tagging; noun phrase identification using Chunking. A preliminary evaluation was conducted to test this technique with the standard IR benchmark collections such as classic test collections and then with the web snippets collection from the search engines results. The experimental results have been encouraging.

Intruder detection by extracting semantic content from surveillance videos

Many surveillance cameras are using everywhere, the videos or images captured by these cameras are still dumped but they are not processed. Many methods are proposed for tracking and detecting the objects in the videos but we need the meaningful content called semantic content from these videos. Detecting Human activity recognition is quite complex. The proposed method called Semantic Content Extraction (SCE) from videos is used to identify the objects and the events present in the video. This model provides useful methodology for intruder detecting systems which provides the behavior and the activities performed by the intruder. Construction of ontology enhances the spatial and temporal relations between the objects or features extracted. Thus proposed system provides a best way for detecting the intruders, thieves and malpractices happening around us.

Valence arousal similarity based recommendation services

Web Services play a vital role in e-commerce and e-business applications. A WS (Web Service) application is interoperable and can work on any platform i.e.; platform independent, large scale distributed systems can be established easily. A Recommender System is a precious tool for providing appropriate recommendations to all users in a Hotel Reservation Website. User based, Top k and profile based approaches are used in collaborative filtering algorithm which does not provide personalized results to the users and inefficiency and scalability problem also occurs due to the increase in the size of large datasets. To address the above mentioned challenges, a Valence-Arousal Similarity based Recommendation Services, called VAS based RS, is proposed. Our proposed mechanism aims to presents a personalized service recommendation list and recommending the most suitable service to the end users. Moreover, it classifies the positive and negative preferences of the users from their reviews to improve the prediction accuracy. For improve its efficiency and scalability in big data environment, VAS based RS is implemented using collaborative filtering algorithm on MapReduce parallel processing paradigm in Hadoop, a widely-adopted distributed computing platform.

Providing Mother and Child Care Telemedicine Through Interactive Voice Response

In India, telemedicine services are provided through satellite communication, Web portals and mobile call centre services. These services enable people to interact with a specialized doctor who is in the other end of the country using videoconferencing and various channels as medium. This saves time and cost of the patient using the service. He/she gets all specialized medical instructions without having the need to see the doctor. This service may be ultimately useful when available to all the people in the country. But, it does not reach everyone in the country. Satellite communication is possible only in places where satellite link is established between the patient and a doctor and Web portals are accessible only by people who have complete knowledge about the existing system mechanism. These methods are mainly available only in urban areas. This is where the problem arises in India. Villages being the backbone of the country and farmers being its soul, they are not aware of these existing technologies and telemedicine application is completely new to them. Providing specialized medical services in cities is easy, but this is not the same for rural India. Therefore, a more efficient, convenient and user-friendly method of implementing telemedicine service is needed. In this paper, we extend this approach of telemedicine to the rural areas using interactive voice response (IVR) system. Since there has been a similar methods already established in urban areas, we are providing this newly proposed system for the rural areas for a better health care.

Prediction of onset of south west monsoon using multiple regression

Indian economy and agriculture is dependent upon the performance of southwest monsoon. In view of the importance of monsoon, a methodology is developed to predict the onset date of southwest monsoon over Kerala, the first entry point of southwest monsoon over Indian main land mass based on the weather data for the period of 1971 to 2012. Here multiple regression method is employed to forecast the onset dates. The data for the period 1971-2000 is used for model development period and the data for 2001-2012 is used to test the model. The model error of multiple regression method is 0.00 for the development period -0.5 for the testing period respectively. The standard deviation (SD) of the actual onset is 6.6039 and our system has error less than this SD and hence proves to be efficient.