Purpose The main contribution of this paper is the development of a methodology for conducting a systematic literature review implemented in a software named SMILE. The methodology and the SMILE tool are applied, as an example, to the field of human resource management in the context of Industry 4.0 or HR4.0. Methods After analysing characteristics of main approaches for systematic literature review (SLRs), we highlighted limitations of methods and tools currently used to support researchers in SLRs and defined an original methodology for automating the SLR phases. The methodology, that can overcome limitations of other methods, has been implemented in a digital tool, SMILE To validate the correctness of the methodology and test the robustness of SMILE, a SLR has been carried out to the case of HR4.0 and main results have been reported. Findings First, a methodology of automatic extraction of the most relevant contributions in the literature is presented. The methodology regards the implementation of Latent Dirichlet Allocation (LDA) as an unsupervised method of topic modelling that enables the identification of relevant topics from a collection of contributions selected from scientific literature. Second, SMILE, a digital tool for conducting systematic literature reviews, has been implemented in the form of a decision support system (DSS). SMILE supports the proposed methodology and is based on machine learning (ML) for natural language processing (NLP) and text analytics technique. Originality/value Compared to the previous review contributions, in this paper the authors propose an automatic methodology for the optimal choice of parameters used by Latent Dirichlet Allocation, such as the optimal number of topics calculated through iterative runs to perform best results in terms of words coherence in topics. SMILE reduces the effort of researchers by the implementation of an interactive procedure that allows the supervision of all the fundamental phases envisaged by the SLR
SMILE: towards an Automated Methodology for Systematic Literature Reviews
A. M. Felicetti;D. Rogano;R. Linzalone
;V. Corvello;S. Ammirato
2021-01-01
Abstract
Purpose The main contribution of this paper is the development of a methodology for conducting a systematic literature review implemented in a software named SMILE. The methodology and the SMILE tool are applied, as an example, to the field of human resource management in the context of Industry 4.0 or HR4.0. Methods After analysing characteristics of main approaches for systematic literature review (SLRs), we highlighted limitations of methods and tools currently used to support researchers in SLRs and defined an original methodology for automating the SLR phases. The methodology, that can overcome limitations of other methods, has been implemented in a digital tool, SMILE To validate the correctness of the methodology and test the robustness of SMILE, a SLR has been carried out to the case of HR4.0 and main results have been reported. Findings First, a methodology of automatic extraction of the most relevant contributions in the literature is presented. The methodology regards the implementation of Latent Dirichlet Allocation (LDA) as an unsupervised method of topic modelling that enables the identification of relevant topics from a collection of contributions selected from scientific literature. Second, SMILE, a digital tool for conducting systematic literature reviews, has been implemented in the form of a decision support system (DSS). SMILE supports the proposed methodology and is based on machine learning (ML) for natural language processing (NLP) and text analytics technique. Originality/value Compared to the previous review contributions, in this paper the authors propose an automatic methodology for the optimal choice of parameters used by Latent Dirichlet Allocation, such as the optimal number of topics calculated through iterative runs to perform best results in terms of words coherence in topics. SMILE reduces the effort of researchers by the implementation of an interactive procedure that allows the supervision of all the fundamental phases envisaged by the SLRI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
