公开(公告)号：JPH10254838A

公开(公告)日：1998-09-25

申请号：JP898398

申请日：1998-01-20

Applicant: IBM

Inventor： PHILIP SHIIRAN YU

IPC: G06F15/16 ,  G06F9/50 ,  G06F12/00 ,  G06F13/00 ,  G06F15/177

Abstract: PROBLEM TO BE SOLVED: To balance a load between distributed sites, which can hold different types of servers, or clustered sites by dynamically calculating the valid interval of a mapping request to the server as the load of the request of a group and one function of server capacity. SOLUTION: The respective servers 54 are the distributed sites or Web servers of the clustered sites. Respective address mapping contains duration (TTL) interval. Standard DNS is improved so that the service request of the client is satisfactorily balanced between the severs 54 of the server group 58. Generalization is executed in extension DNS 62. Extension DNS 62 maps the name request to an IP address and calculates TTL as the gateway request load and (or) the function of the capacity of a destination server associated with a IP address.

公开(公告)号：JP2001339517A

公开(公告)日：2001-12-07

申请号：JP2001080630

申请日：2001-03-21

Applicant: IBM

Inventor： MOSKOWITZ PAUL A ,  PHILIP SHIIRAN YU ,  DINKIN SAM ,  BOIES STEPHEN J

IPC: G06F17/28 ,  G06F15/00 ,  G10L13/00 ,  G10L13/04 ,  G10L15/00 ,  H04M3/42 ,  H04M3/493 ,  H04W88/18 ,  H04Q7/38

Abstract: PROBLEM TO BE SOLVED: To provide a system and method for information conversion that facilitates exchange of many digital information data between wireless communication units in order to aid communication between users understanding different languages. SOLUTION: This invention provides a system and a method for information conversion employing wireless communication units. The communication unit allows a user to enter data in its mother language. The communication unit converts an entry request into a markup language. The communication unit transmits a markup language expressing a property of a request and a mother language of the user of the communication unit to a 2nd communication unit via a wireless link. The 2nd communication unit receives the request and returns a reply in a mother language instructed by the user of the 1st communication unit or in the markup language. The personal communication unit receives the reply and displays the reply in the mother language of the original user.

公开(公告)号：JP2001273186A

公开(公告)日：2001-10-05

申请号：JP2001015963

申请日：2001-01-24

Applicant: IBM

Inventor： WU KUN-LUNG ,  PHILIP SHIIRAN YU

IPC: G06F12/00 ,  G06F13/00 ,  G06F17/30

Abstract: PROBLEM TO BE SOLVED: To provide hashing corresponding to a waiting time to common Web caching between proxy caches distributed geographically. SOLUTION: This method performs common Web caching between the proxy caches distributed geographically and, in particular this system/method for waiting time-corresponding hashing performs common Web caching between the proxy caches distributed geographically. In hashing, the delay of network waiting time and a proxy loading state are considered. When the loading state is good as the result, a request can be hashed to the proxy caches which is geographically close. in other cases, the request is hashed to geographically remote proxy caches to balance the load between the caches.

公开(公告)号：JPH10240770A

公开(公告)日：1998-09-11

申请号：JP3450998

申请日：1998-02-17

Applicant: IBM

Inventor： PHILIP SHIIRAN YU ,  BOB CHAO-CHU LIAN ,  MIN-SHAN CHEN

IPC: G06F9/44 ,  G06F17/30 ,  G06N5/02 ,  G06N5/04 ,  G06Q30/02

Abstract: PROBLEM TO BE SOLVED: To retrieve a causality principle rule from an event database wherein respective events can belong to more than one category by generating candidate rules for the retrieval of the causality principle rule, and repeating a rule that counts their generation. SOLUTION: A processor 101 applies an causality principle retrieval application 106 to information retrieved from a permanent storage position 102 by using a memory buffer 105. For example, a transaction database is so constituted that a 1st sequence includes an event Gi by performing preprocessing so as to generate an event sequence data base of N sequences. The preprocessing is performed by carrying out a classification phase wherein transaction records are classified according to customer IDs and time and a mapping phase wherein respective transaction events are mapped to relative categories. For the purpose, hierarchical subsequence matching is used and a candidate causality principle rule having the set size of a result increasing gradually for the matching is repeatedly generated.