 | Saxena, Anurag | (2) |
 |
 | Exploring models for the growth of literature data
Author :Saxena, Anurag;Gupta, B.M.;Jauhari, Monika
Source: DESIDOC Bulletin of Information Technology :27(3) ; 2007 ; pp.03-12
Abstract : Any time series is forecasted using a suitable model based on the analysis of historical data. The value of a model lies in the efficacy with which it performs the task for which it has been constructed. A model is considered good if it fits the data well. In other words, models should have good parameter values and fit statistics. Many researchers have successfully applied various statistical models to analyse the growth of literature data. However, there is no generalised rule or procedure put forwarded by these researchers. The question, therefore, arises as to how one compares the appropriateness of different type of models fitted to the data? The aim of this paper is to forecast the time series of growth of literature data. Two different approaches to probe this kind of data have been applied. These approaches are the multiplicative seasonal model approach and nonlinear model approach where the trend has an exponential growth form. It has been shown that there is plethora of models that come out with good fit parameters. This communication thus highlights some basic issues related to forecast of growth of literature data.
Keywords : Historical data;Time senses;Dynamic models |
| | Zipf’s law in a random text from english with a new ranking method
Author :Saxena, Anurag;Jauhari, Monika;Gupta, B.M.
Source: DESIDOC Bulletin of Information Technology :27(4) ; 2007 ; pp.51-58
Abstract : Zipf’s law has attracted infometricians time and again. There have been many studies, which have explored the application of Zipf’s law to various areas. However, there are a few parameters, which largely affect a study. These parameters are the power law embedded in Zipf’s law, the ranking method, the type of text taken for the study and the behaviour of extreme regions in the Zipf’s curve. This paper tries to address all these points by taking a random text in English language from computer science literature. The selected text is called random because of its highly specific nature of technical words. The paper studies the properties of this text and compares the product of rank and frequency for three ranking procedures. It also analyses the performance of data in the extreme regions of the Zipf’s curve. It is observed that ranking procedure and type of text have definite bearings on the performance of Zipf’s curve.
Keywords : Zipf’s law;Infometrics;Zipf’s curve;Power law;Computer science |
|
