This paper proposes a hybrid of handcrafted rules and a machine learning method for chunking Korean. In the partially free word-order languages such as Korean and Japanese, a small number of rules dominate the performance due to their well-developed postpositions and endings. Thus, the proposed method is primarily based on the rules, and then the residual errors are corrected by adopting a memo...

This paper presents an on-going effort which aims to annotate the Wall Street Journal sections of the Penn Treebank with the help of a hand-written large-scale and wide-coverage grammar of English. In doing so, we are not only focusing on the various stages of the semi-automated annotation process we have adopted, but we are also showing that rich linguistic annotations, which can apart from sy...

Person re-identification (re-ID) aims to accurately retrieve a person from a large-scale database of images captured across multiple cameras. Existing works learn deep representations using a large training subset of unique persons. However, identifying unseen persons is critical for a good re-ID algorithm. Moreover, the misalignment between person crops to detection errors or pose variations l...

Evidence is provided that players pay at least as much attention to a set of procedurally generated Sokoban levels as they do to levels hand crafted by expert designers. Data were collected from 40 participants who played Sokoban under laboratory conditions while simultaneously performing an auditory Stroop test. Three performance measures from the Stroop test were analyzed and compared after a...

The 200 million word corpus of the Bank of English was annotated morphologically and syntactically using the English Constraint Grammar analyser, a rulebased shallow parser developed at the Research Unit for Computational Linguistics, University of Helsinki. We discuss the annotation system and methods used in the corpus work, as well as the theoretical assumptions of the Constraint Grammar syn...

The automated generation of playlists represents a particular type of the music recommendation problem with two special characteristics. First, the tracks of the list are usually consumed immediately at recommendation time; second, tracks are listened to mostly in consecutive order so that the sequence of the recommended tracks can be relevant. A number of different approaches for playlist gene...

This paper is a review of a story generation system called Minstrel. It uses complex but hand-crafted Lisp knowledge structures to generate short computer-generated stories within the King Arthur domain. The knowledge representation model of Minstrel is reimplemented using a W3C standard language to analyze the pros and cons of technology updates over this kind of classic AI projects.

Partially Observable Markov Decision Processes (POMDPs) are attractive for dialogue management because they are made to deal with noise and partial information. This paper addresses the problem of using them in a practical development cycle. We apply factored POMDP models to three applications. We examine our experiences with respect to design choices and issues, and compare performance with ha...

This paper examines how far it is possible to replace hand crafted dialogue design with a combination of generic dialogue system components and ontological domain knowledge. Two case studies are presented, one for dialogue-based breast cancer referrals, the other for control of networked home appliances.

We use evolutionary computation techniques to create real-time reactive controllers for a race-car simulation game: RARS (Robot Auto Racing Simulator). Using genetic programming to evolve driver controllers, we create highly generalized game-playing agents, able to outperform most human-crafted controllers and all machine-designed ones on a variety of game tracks.