Update academicpages#6 - publication fixed

_publications/2022-08-15-pace.md

@@ -1,5 +1,5 @@
 ---
-title: "Using Domain Knowledge to Assist Process Scenario Discoveries"
+title: "PACE: Preventing Attacks on Case Identities in Event Logs Through Attribute Generalizations"
 collection: publications
 permalink: /publication/2022-08-15-pace
 excerpt: 'Third Author'
@@ -8,4 +8,7 @@ venue: '2022 IEEE 46th Annual Computers, Software, and Applications Conference (
 paperurl: 'https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9842728'
 citation: 'Zhang, Zhu, J., Hildebrant, R., Venkatasubramanian, N., & Ren, S. (2022). Using Domain Knowledge to Assist Process Scenario Discoveries. In 2022 IEEE 46th Annual Computers, Software, and Applications Conference (COMPSAC) (pp. 226–288). IEEE. https://doi.org/10.1109/COMPSAC54236.2022.00047'
 ---
-In this Research-In-Progress work, we present a potentially generalizable approach that can determine the sensitivity of an attribute or a set of attributes associated with an event in a given event log. This approach is based on the concept of an equivalence class that a given event or trace may form and associate its sensitivity with the size of its equivalence class. For a given event, different equivalent classes can be formed based on different attributes, the proposed approach provides researchers with a more granular tool to apply group based privacy to event logs.
+Process Mining is an emerging research field that looks at event logs to build graphical models and provides new insights to businesses that allow them to make process-driven decisions. While there are many benefits to process mining, some businesses and researchers have hesitations about adopting process mining in real applications because of sensitive attribute data contained in an event log. To deal with this issue, researchers have developed tools and frameworks that apply privacy to event-logs. In their work, they only consider attacking privacy from a control-flow perspective and do not fully address potential privacy leakages that can be created from attributes. In PACE, we introduce a privacy-enhancing framework that focuses on the generalization of attribute values based on different organizational perspectives. This privacy framework comprises three components: control-flow anonymization, heuristic-driven hierarchy selection for anonymizing attributes, and application of attribute generalizations based on a perspective. To assess our framework, we apply PACE to the BPIC 2013 Event Log and measure the retained precision of handovers, the effect of the logs on decision trees, and show a sensitivity analysis of our privacy logs. Additionally, we show that PACE’s results greatly outperforms a state-of-the-art differential privacy tool on the same organization mining tasks.
+
+
+