How to find the needle in a haystack : Can computers help authors to overcome the tip-of-the tongue state ?
Speaker： Dr. Michael Zock
(CNRS-LIF, Aix-Marseille Université, France)
時間：2/19 13:30
地點：台達館 105
講題：
How to find the needle in a haystack :
Can computers help authors to overcome the tip-of-the tongue state ?
摘要:
We spend a large amount of our lives searching : ideas, words, CDs, documents and “you just name it”. Our success of finding the desired object depends to a large extend on our abilities to put it in the right place and to describe it in adequate terms. Indexing is a skill, and like good categorization it is prerequisite for efficient storage and I will illustrate this problem for word access. More precisely, I will show how electronic dictionaries could be enhanced to help authors to find the word they are looking for (something most, if not all, dictionaries are still very poor at). The underlying rational is twofold : we need to build a resource within which search takes place, and we need to provide good exploratory tools, i.e. means for easy acces (search or navigation). The solution to the first problem is an association-based index. The solution to the second problem is clustering. I will discuss in my talk what currently exists, what is needed and what the bottlenecks are, and I will also try to show that the solution, while not easy, is within reach.
(CNRS-LIF, Aix-Marseille Université, France)
時間：2/19 13:30
地點：台達館 105
講題：
How to find the needle in a haystack :
Can computers help authors to overcome the tip-of-the tongue state ?
摘要:
We spend a large amount of our lives searching : ideas, words, CDs, documents and “you just name it”. Our success of finding the desired object depends to a large extend on our abilities to put it in the right place and to describe it in adequate terms. Indexing is a skill, and like good categorization it is prerequisite for efficient storage and I will illustrate this problem for word access. More precisely, I will show how electronic dictionaries could be enhanced to help authors to find the word they are looking for (something most, if not all, dictionaries are still very poor at). The underlying rational is twofold : we need to build a resource within which search takes place, and we need to provide good exploratory tools, i.e. means for easy acces (search or navigation). The solution to the first problem is an association-based index. The solution to the second problem is clustering. I will discuss in my talk what currently exists, what is needed and what the bottlenecks are, and I will also try to show that the solution, while not easy, is within reach.