Rajib K. Das

Unsupervised Learning of Morphology for Building Lexicon for a Highly Inflectional Language

Words play a crucial role in aspects of natural language understanding such as syntactic and semantic processing. Usually, a natural language understanding system either already knows the words that appear in the text, or is able to automatically learn relevant information about a word upon encountering it. Usually, a capable system---human or machine, knows a subset of the entire vocabulary of a language and morphological rules to determine attributes of words not seen before. Developing a knowledge base of legal words and morphological rules is an important task in computational linguistics. In this paper, we describe initial experiments following an approach based on unsupervised learning of morphology from a text corpus, especially developed for this purpose. It is a method for conveniently creating a dictionary and a morphology rule base, and is, especially suitable for highly inflectional languages like Assamese. Assamese is a major Indian language of the Indic branch of the Indo-European family of languages. It is used by around 15 million people.

Acquisition of Morphology of an Indic Language from Text Corpus

This article describes an approach to unsupervised learning ofmorphology from an unannotated corpus for a highly inflectionalIndo-European language called Assamese spoken by about 30 millionpeople. Although Assamese is one of Indias national languages, itutterly lacks computational linguistic resources. There exists noprior computational work on this language spoken widely innortheast India. The work presented is pioneering in this respect.In this article, we discuss salient issues in Assamese morphologywhere the presence of a large number of suffixal determiners,sandhi, samas, and the propensity to use suffix sequences makeapproximately 50% of the words used in written and spoken textinflected. We implement methods proposed by Gaussier and Goldsmithon acquisition of morphological knowledge, and obtain F-measureperformance below 60%. This motivates us to present a method moresuitable for handling suffix sequences, enabling us to increase theF-measure performance of morphology acquisition to almost 70%. Wedescribe how we build a morphological dictionary for Assamese fromthe text corpus. Using the morphological knowledge acquired and themorphological dictionary, we are able to process small chunks ofdata at a time as well as a large corpus. We achieve approximately85% precision and recall during the analysis of small chunks ofcoherent text.

Diameter and Routing in Enhanced OTIS Cube

Enhanced OTIS-cube (E-OTIS-Q_n), a variation of the OTIS-cube (OTIS-Q_n) was proposed in (R.K. Das, 2005). E-OTIS-Q_n is regular of degree n+1 and is obtained from the normal OTIS-cube by adding some extra links. In (R.K. Das, 2005), it was shown that the diameter of E-OTIS-Q_n is less than or equal to lfloor4n+5/3rfloor and a heuristic for point-to-point routing has been proposed. In this paper, an optimal algorithm for one-to-one routing in E-OTIS-Q_n has been developed. We have shown that the diameter of E-OTIS-Q_n is equal to lfloor4n+4/3rfloor which is almost two-third of the diameter of OTIS-Q_n

Adaptive fault tolerant routing in star graph

In this paper a fault tolerant routing algorithm on star graph is proposed. Each node in an n-star is associated with a fault-vector of d=

Power Efficient Data Gathering for Sensor Network

\lceil \frac{3(n-1)}{2} \rceil

Heuristic-Based Resource Reservation Strategies for Public Cloud

bits, which is an approximate measure of the number and distribution of faults in the neighborhood. The routing algorithm based on the fault-vector finds the shortest path between any source-destination pair in presence of large number of faulty nodes or links. Simulation results show that the % of cases where the algorithm fails to find a shortest path even when a shortest path exists, is as low as 2.7 with 140 node or link faults in a 7-star.

Target coverage using a collaborative platform for sensor cloud

In this paper we have presented an algorithm to construct a rooted tree with base station as root connecting all the sensor nodes. The tree is constructed with the aim of maximizing the network life-time. It is assumed that all the nodes have same initial energy, but they can adjust their transmission range and thus the amount of energy needed for transmission may vary. The cost of a node is the amount of energy spent by it in each data gathering round. While determining the lifetime of a node, the energy lost in the process of constructing the tree is also considered. Thus the lifetime of a node is its residual energy (initial energy minus energy spent in exchanging messages for construction) divided by its cost. The lifetime of the network is the minimum of the lifetimes of all the nodes in the sensor network. It is also assumed the sensed data is aggregated so that nodes send a fixed sized message to its parent in each data gathering round. The algorithm works in two phases: In the first phase, an initial tree is constructed where the path from a sensor node to the base station consists of least possible number of hops. In the second phase (called fine-tuning) nodes may change their parents if that lead to a reduction in maximum cost of the three nodes involved (the node, its present parent, its future parent). The algorithms for both the phases (initial tree construction and fine-tuning) are distributed where nodes take decision Bbased on the status of its neighbors. Experimental results show that fine-tuning leads to considerable improvement in life-time of the network. The lifetime computed is significantly higher than those obtained by other well-known algorithms for data gathering.

Finding the maximum lifetime data-gathering tree in sensor networks

Cloud service providers (CSPs) adapt different pricing models for their offered services. Some of the models are suitable for short term requirement while others may be suitable for the cloud service users (CSU) long term requirement. For example, reservation-based pricing model is appropriate for a CSUs long term demand for resources. Finding the optimal amount of resources to be reserved in advance, to minimize the total cost, needs sufficient research effort. Various algorithms were discussed in the last couple of years to solve the resource reservation problem but most of them are based on integer programming problem (IPP) which is NP in nature. In this paper, we derive some heuristic-based polynomial time algorithms to find some near optimal solution to this problem. We show that the cost for CSU using our approach is comparable to the solution obtained using optimal IPP.

Fault tolerant routing using improved safety vectors

Nowadays Sensor-cloud has received a lot of interests among researchers in the field of Wireless Sensor Networks (WSN) as well as Cloud Computing. An integral part of the Sensor Cloud platform is the Collaboration of WSN providers which improves the utilization of sensor resources. The physical sensors in such a collaborative environment are shared among multiple applications from different administrative domains. In this paper, we propose a Sensor Cloud Architecture which allows the users to deploy their sensor applications on-demand within the collaborative platform. The sensor resources used by the application are provided as a service to the user. The cost for offering such service depends on the number of physical sensors selected for covering the targets (locations of interest) for that application as well as connecting the selected physical sensors to the base station. In this paper, we proposed a two phase algorithm for this problem. In the first phase (Coverage), an Integer Programming Problem(IPP) is used to find a minimum set of sensors to cover the targets. In the next phase (Connectivity) a heuristic is used to extend the set such that connectivity of all the selected sensors is ensured. The performance of the proposed approach is validated through simulation which shows significant improvement of using collaboration over multiple non-collaborative WSNs. We also establish the superiority of the heuristic for ensuring connectivity over one such algorithm in the literature.

Fault tolerant routing in star graphs using fault vector

In this work, we study the construction of a data gathering tree to maximize the network lifetime, which is defined as the minimum of the lifetime of all the nodes in the sensor network. The problem of finding the optimal data gathering tree is known to be NP-complete [9]. Most works in this area aim to give an heuristic to find a data gathering tree and compare the lifetime obtained with other existing solutions. But it is beyond doubt that best tree with respect to lifetime could be obtained if we can generate all possible spanning trees and choose among them the one giving maximum lifetime. With a sensor network consisting of large number of nodes this approach becomes computationally prohibitive. We have adopted an algorithm to generate all possible spanning trees and modified it to generate only the optimal spanning tree using a branch and bound technique. Here, whenever a partial tree has a node with lifetime less than or equal to the lifetime of a tree generated earlier, that tree is discarded. Simulation results show that the number of complete trees checked by our method is only a small fraction of the total number of possible spanning trees. That way it becomes feasible to find a solution in reasonable time even for large sensor network. We have considered that transmission range of each sensor can be adjusted to be just enough to send a message to its parent. The lifetime obtained by our method is also shown to be better than that obtained by other heuristic algorithm like PEDAP.