We study the consistency of listwise ranking methods with respect to the popular Normalized Discounted Cumulative Gain (NDCG) criterion. State of the art listwise approaches replace NDCG with a surrogate loss that is easier to optimize. We characterize NDCG consistency of surrogate losses to discover a surprising fact: several commonly used surrogates are NDCG inconsistent. We then show how to ...

Using Yelp data, we built a restaurant recommendation system for individuals and groups. For each of 4.9k individual Yelp users, we create a ranking SVM model with features encompassing users’ food preferences and dietary restrictions, such as cuisine type, services o↵ered, ambience, noise level, average rating, etc. Our recommendation system for individual users achieved 72% average prediction...

Existing approaches on XML keyword search mostly focus on querying over single data source. However, searching over hundreds or even thousands of (distributed) data sources by sequentially querying every single data source is extremely high cost, thus it can be impractical. In this paper, we propose an approach for selecting top-k data sources to a given query in order to avoid high cost of sea...

Suppose all the individuals in a field are linearly ordered. Groups of individuals form teams. Is there a perfect ranking function of each team based on the members of the team? We prove that under a very mild and reasonable set of axioms for the ranking function, no such ranking function exists. AMS Subject Classification: 68R05, 91F10, 06A05, 06A07.

Much progress has been made in the English question-answering task since it was initiated in TREC-8 and our last participation in TREC-2001. QA is a complex semantics-oriented task, and it is necessary that much linguistic processing, auxiliary resources, and learning steps are needed to come up with adequate performance to the task [1]. Chinese language factoid QA was first introduced during N...

For some multi-criteria decision aiding methods, the relative ranks of two actions may be inverted when the original set is altered. This phenomenon is known as rank reversal. In this contribution, we formalise rank reversal for the Promethee II method and derive the exact conditions for its occurrence when one ore more actions are added or removed from/to the original set. These conditions eve...

Distributions over rankings are used to model data in various settings such as preference analysis and political elections. The factorial size of the space of rankings, however, typically forces one to make structural assumptions, such as smoothness, sparsity, or probabilistic independence about these underlying distributions. We approach the modeling problem from the computational principle th...

Traditional mechanisms used to rank learning objects are not longer viable thanks to the current abundance of resources. This work proposes the construction of an improved relevance ranking function based on contextualized attention data extracted from the interaction between the users and the objects. A multidimensional approach to relevance is followed. Beside the Algorithmic relevance, curre...

BACKGROUND Mexican-American fertility is poorly understood because data limitations prevent researchers from accurately estimating the fertility levels of members of this group and from determining how their fertility changes within and across generations. OBJECTIVE Using binational data and an innovative methodological addressing key methodological limitations, I (1) estimate the fertility of ...