Systematising the LCA approaches’ soup: a framework based on text mining

Purpose: It is challenging for practitioners to navigate through the multitude of life cycle assessment (LCA) approaches due to the rich literature and a lack of systematisation. The LCA flexibility allowed by standards results in a multitude of applications and, as referred to in previous works, as an “alphabet soup”. This paper proposes a scheme for a clearer classification of currently used LCA approaches, with consideration of the 4-stage framework coming from standards. Methods: This systematisation was first established through literature research serving as a preliminary tentative framework. A text mining task was carried out in a second stage, involving 2044 published articles among 7558 of the last 10 years. For text mining, a dictionary collected keywords and synonyms of the LCA approaches. Such keywords were then extracted from the text together with their context (multiword). The final multiword analysis allowed the association of each keyword (i.e. each LCA approach) with a specific LCA stage (Goal and Scope, Life Cycle Inventory, Life Cycle Impact Assessment, Interpretation). The preliminary framework was adapted, further enriched and validated based on the text mining results. Results: As a result of the text mining activities, the preliminary tentative framework was partially confirmed and enriched with new insights, especially in the field of “explorative” LCA approaches, which also include “prospective” and “scenario-based” LCA. For most of the currently used LCA approaches, a link to a unique LCA stage was not recorded. However, clear trends were detected. The text mining task also highlighted a high number of works in which different approaches are compared or counterposed, especially in the field of attributional and consequential LCA. Some issues were found with the connotations of “traditional” approaches, which could be defined more specifically as “non-explorative”. Conclusions: Unlike other works focused on notions from selected literature, text mining activities can provide bottom-up feedback on a larger scale more automatically. In addition, this work brought out novel LCA approaches, for which future developments will confirm a final definition and systematisation. As an additional advantage, the presented methodology is easily replicable. Hence, the presented framework can be updated along with developments in LCA approaches.