Sarwat Nizamani

From public outrage to the burst of public violence: An epidemic-like model

This study extends classical models of spreading epidemics to describe the phenomenon of contagious public outrage, which eventually leads to the spread of violence following a disclosure of some unpopular political decisions and/or activity. Accordingly, a mathematical model is proposed to simulate from the start, the internal dynamics by which an external event is turned into internal violence within a population. Five kinds of agents are considered: “Upset” (U), “Violent” (V), “Sensitive” (S), “Immune” (I), and “Relaxed” (R), leading to a set of ordinary differential equations, which in turn yield the dynamics of spreading of each type of agents among the population. The process is stopped with the deactivation of the associated issue. Conditions coinciding with a twofold spreading of public violence are singled out. The results shed new light to understand terror activity and provides some hint on how to curb the spreading of violence within population globally sensitive to specific world issues. Recent violent events in the world are discussed.

CCM: A Text Classification Model by Clustering

In this paper, a new Cluster based Classification Model (CCM) for suspicious email detection and other text classification tasks, is presented. Comparative experiments of the proposed model against traditional classification models and the boosting algorithm are also discussed. Experimental results show that the CCM outperforms traditional classification models as well as the boosting algorithm for the task of suspicious email detection on terrorism domain email dataset and topic categorization on the Reuters-21578 and 20 Newsgroups datasets. The overall finding is that applying a cluster based approach to text classification tasks simplifies the model and at the same time increases the accuracy.

Semantic analysis of FBI news reports

In this paper we present our work on semantic analysis of FBI News reports. In the paper we have considered the News which are of the immense significance for the analyst who want to analyze the News of specific area. With this definite analysis we are able to extract critical events or concepts described in News along with entities involved in the event. These entities include important actors of the event or concept, with location and temporal information. This information will help News analyzers to retrieve the information of interest efficiently.

Testing and validating the ERP success evaluation model for higher education institutes of Pakistan

Higher Education Commission (HEC) of Pakistan has taken profound measures to bring significant improvement in universities of Pakistan. HEC signed an agreement with Oracle PeopleSoft to equip universities with high-tech information technology (IT) solutions. To this end, HEC initially implemented enterprise resource planning (ERP) campus management system in seven universities of Pakistan as a pilot project. However, there is a void in literature to evaluate these ERP implementations. This research formulates a conceptual model based on strong background theories to effectively evaluate the success of ERP system implementation. A questionnaire based online survey approach is used to collect data from 323 respondents. Findings of the paper result in a refined model for analysing success of ERP implementation. By employing this model, the paper also reports on the success and failure of ERP implementation. Finally, the undertaken study implies that its findings may substantially add regarding the future decisions of ERP implementations.

PSN: Portfolio Social Network

 Abstract—In this paper we present a web-based information system which is a portfolio social network (PSN) that provides solutions to recruiters and job seekers. The proposed system enables users to create portfolios so that he/she can add his specializations with piece of code, if any, specifically for software engineers, which is accessible online. The unique feature of the system is to enable the recruiters to quickly view the prominent skills of the users. A comparative analysis of the proposed system with the state of the art systems is presented. The comparative study reveals that the proposed system has advanced functionalities.

Evolution of Terrorist Network Using Clustered Approach: A Case Study

In the paper we present a cluster based approach for terrorist network evolution. We have applied hierarchical agglomerative clustering approach to 9/11 case study. We show that, how individual actors who are initially isolated from each other are converted in small clusters and result in a fully evolved network. This method of network evolution can help intelligence security analysts to understand the structure of the network.

Detection of Illegitimate Emails Using Boosting Algorithm

In this paper, we report on experiments to detect illegitimate emails using boosting algorithm. We call an email illegitimate if it is not useful for the receiver or for the society. We have divided the problem into two major areas of illegitimate email detection: suspicious email detection and spam email detection. For our desired task, we have applied a boosting technique. With the use of boosting we can achieve high accuracy of traditional classification algorithms. When using boosting one has to choose a suitable weak learner as well as the number of boosting iterations. In this paper, we propose suitable weak learners and parameter settings for the boosting algorithm for the desired task. We have initially analyzed the problem using base learners. Then we have applied boosting algorithm with suitable weak learners and parameter settings such as the number of boosting iterations. We propose a Naive Bayes classifier as a suitable weak learner for the boosting algorithm. It achieves maximum performance with very few boosting iterations.

Detecting Terrorism Incidence Type from News Summary

The paper presents the experiments to detect terrorism incidence type from news summary data. We have applied classification techniques on news summary data to analyze the incidence and detect the type of incidence. A number of experiments are conducted using various classification algorithms and results show that a simple decision tree classifier can learn incidence type with satisfactory results from news data.

Cluster Based Text Classification Model

We propose a cluster based classification model for suspicious email detection and other text classification tasks. The text classification tasks comprise many training examples that require a complex classification model. Using clusters for classification makes the model simpler and increases the accuracy at the same time. The test example is classified using simpler and smaller model. The training examples in a particular cluster share the common vocabulary. At the time of clustering, we do not take into account the labels of the training examples. After the clusters have been created, the classifier is trained on each cluster having reduced dimensionality and less number of examples. The experimental results show that the proposed model outperforms the existing classification models for the task of suspicious email detection and topic categorization on the Reuters-21578 and 20 Newsgroups datasets. Our model also outperforms A Decision Cluster Classification (ADCC) and the Decision Cluster Forest Classification (DCFC) models on the Reuters-21578 dataset.