Dawei Wang, PhD Student
1. E+TRA (Electronic TRAnsparency & tracking)
A web-based, multi-platform, centralized, offline-compatible electronic emergency response system
Motivation: The efficiency of humanitarian aid supply chains is critical to effective emergency response. In addition, a sustainable supply chain network promotes the recovery of devastated areas and the affected populations. Thus, an efficient, reliable and sustainable emergency supply chain management system plays a critical role in the impact of humanitarian relief. Currently, a typical NGO ERP (enterprise resource planning) system does not address the specific needs of emergency humanitarian relief functions. Without an emergency supply chain management system, the lack of coordination and potential human errors will cause delays and waste in resources, which jeopardizes the recovery and even the survival of affected people. In addition, tracking items from donor resources all the way to distribution to target beneficiaries is important for accountability and transparency.
Description: This system is capable of connecting all global warehouses across different country programs, requesting and approving relief materials, checking inventory levels, tracking relief materials from donors to beneficiaries, and automatically generating accounting and beneficiary reports. This application will help NGOs form efficient, reliable and sustainable distribution plans for emergency responses with the most updated information.
Result: IEEE GHTC 2016 paper: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7857288
Video: Contact yih@purdue.edu
Also presented at ICT4D 2017, Logistic Cluster 2018
Collaborators: CRS, funders: I2D lab, CRS, Regenstrief
2. Time-aware collaborative filtering recommendation system
Motivation: A product or service is not judged only by its own characteristics but also by the characteristics of other products or services offered concurrently or by anchoring based on users memory. Rating or satisfaction is viewed as a function of the discrepancy or contrast between obtained and expected outcomes. This is documented as contrast effects [36]. Thus, the score of rating can be affected by the sequence of rating. However, in traditional collaborative filtering, pairwise similarities measured between items do not consider the sequence of rating, which could introduce biases caused by contrast effects. Moreover, It is well known that consumers’ preferences can shift over time. This can be due to (1) exploration of new items, instead of repeatedly interacting with the same items. New items may continuously enter the market as well. (2) experience or tendency to interact with items they have previously had positive interactions with, and to stop interacting with items they have previously negative interactions with. (3) popularity of the items, irrespective of personal interaction history [46]. (4) social influence may cause preference changes as a result of observing the preference changes of friends.
Description: Memory-based collaborative filtering is one of the recommendation system methods used to predict a user’s rating or preference by exploring historic ratings, but without incorporating any content information about users or items. It can be either item-based or user-based. Taking item-based Collaborative Filtering (CF) as an example, the way it makes predictions is accomplished in 2 steps: first, it selects based on pair-wise similarities a number of most similar items to the predicting item from those that the user has already rated on. Second, it aggregates the user’s opinions on those most similar items to predict a rating on the predicting item. Thus, similarity measurement determines which items are similar, and plays an important role on how accurate the predictions are. Many studies have been conducted on memory-based CFs to improve prediction accuracy, but none of them have achieved better prediction accuracy than state- of-the-art model-based CFs. we proposed a new approach that combines both structural and rating-based similarity measurement. We found that memory-based CF using combined similarity measurement can achieve bet- ter prediction accuracy than model-based CFs in terms of lower MAE and reduce memory and time by using less neighbors than traditional memory-based CFs on MovieLens and Netflix datasets. Furthermore, we incorporate user preference dynamics into collaborative filtering by modifying similarity measurements and user preference shifts between categories.
Result: One paper under view submitted to Information Science
Collaborators: Mario Ventrestca