Discovery of Activities’ Actor Perspective from Emails based on Speech Acts Detection

Multi‐perspective business process discovery from messaging systems: State‐of‐the art

“…14 The second sub-category regroups the approaches that enable the discovery of multiple activities per message. [15][16][17][18][19][20][21][22][23]…”

“…To enable the discovery of multiple activities per message, proposed approaches relies on : (1) splitting messages into sentences and then, assigning these sentences to activities (using supervised or unsupervised learning techniques), [16][17][18][19][20][21] or (2) associating to each activity (manually or automatically) a set of patterns of words enabling its detection in messages without being constrained by messages or sentences structures. 15,22,23 Phrases based approaches: These approaches suppose that one activity is expressed at a sentence level. In Figure 5, we illustrate the main steps adopted by them to discover activities from messages.…”

“…Recently, Elleuch et al 22,23 discovered activities as well as their actor perspective on the bases of patterns and speech acts detection from emails.…”

“…These works 22,23 have the advantage of proposing completely non supervised approaches, which facilitate knowledge extraction without human intervention. Additionally, compared to previous works, new perspectives (data and actors ones) were discovered and studied.…”

“…Only Van der Aalst et al 30 For activity discovery approaches, some of them proposed to handle data and organizational perspectives in addition to the functional perspective. [20][21][22][23] Two works 20,21 have partially studied them. They discovered data perspective by detecting some predefined data types such as attached documents and email-included web pages descriptions.…”