Particle Swarm Model Selection for Authorship Verification

Abstract

Abstract. Authorship verification is the task of determining whether documents were or were not written by a certain author. The problem has been faced by using binary classifiers, one per author, that make individual yes/no decisions about the authorship condition of documents. Traditionally, the same learning algorithm is used when building the classifiers of the considered authors. However, the individual problems that such classifiers face are different for distinct authors, thus using a single algorithm may lead to unsatisfactory results. This paper describes the application of particle swarm model selection (PSMS) to the problem of authorship verification. PSMS selects an ad-hoc classifier for each author in a fully automatic way; additionally, PSMS also chooses preprocessing and feature selection methods. Experimental results on two collections give evidence that classifiers selected with PSMS are advantageous over selecting the same classifier for all of the authors involved.

Ensemble particle swarm model selection

“…Therefore, PSMS solutions are combinations of such techniques with different parameters settings; a sample (decodified) solution in PSMS is as follows: Under the above model the data is first standardized, next the s2n feature selection method is used for selecting at most f = 8 features, then a neural classifier with specific parameters is used for classification. PSMS has reported satisfactory results on diverse binary classification problems without requiring significant supervision [5], [7], [4]. The main benefits of PSMS is that (1) very effective models can be obtained, (2) no knowledge is required on machine learning nor on the application domain and (3) it can be applied to any binary classification problem.…”

“…the best classification model) is selected. Satisfactory results have been reported with PSMS on a variety of domains [6], [5], [7]. However, many of the evaluated models that are potentially useful for the classification problem are disregarded for the final model.…”

Term-weighting learning via genetic programming for text classification

“…Although, TC is a widely studied topic with very important developments in the last two decades (Sebastiani, 2008;Feldman and Sanger, 2006), it is somewhat surprising that little attention has been paid to the development of new TWSs to better represent the content of documents for TC. In fact, it is quite common in TC systems that researchers use one or two common TWSs (e.g., B, TF or TF-IDF ) and put more effort in other processes, like feature selection (Forman, 2003;Yang and Pedersen, 1997), or the learning process itself (Agarwal and Mittal, 2014;Aggarwal, 2012;Escalante et al, 2009). Although all of the TC phases are equally important, we think that by putting more emphasis on defining or learning effective TWSs we can achieve substantial improvements in TC performance.…”

Improving the Reproducibility of PAN’s Shared Tasks:

“…Previous work on author verification has been evaluated using sample texts in one language only (Greek [60], Dutch [17,30], English [25,26]) and a specific genre (newspaper articles [60], student essays [30], fiction [25], newswire stories [19], poems [19], blogs [26]). Author verification was also included in previous editions of PAN: the author identification task at PAN-2011 included three author verification problems [1], PAN-2013 focused on author verification and provided corpora in English, Greek, and Spanish [22].…”

“…A variety of performance measures have been used in previous work on this task including false acceptance and false rejection rates [60,17], accuracy [25,26], recall, precision, F 1 [30], balanced error rate [19], recall-precision graphs [26] macro-average precision and recall [1], and ROC graphs [22]. Unfortunately, these measures are not able to explicitly estimate the ability of an approach to leave problems unanswered-a fact which is crucial in a cost-sensitive task like this.…”