Bài giảng Hệ cơ sở tri thức (Tuần 11)

9/10/2013

Tuần 11 (Week 11)

Hai V Pham hai@spice.ci.ritsumei.ac.jp 1

•KKKKỹ thuthuthuthuậtttt thuthuthuthu ththththậpppp tri chuyên mônmônmônmôn tri ththththứcccc chuyên chuyên chuyên tri tri ttttừ chuyên chuyên mônmônmônmôn chuyên giagiagiagia vàvàvàvà tàitàitàitài lilililiệuuuu chuyên chuyên chuyên chuyên chuyên •ÁpÁpÁpÁp ddddụngngngng bàibàibàibài ttttậpppp llllớnnnn

(cid:1)Tham khảo tài liệu trong nước, quốc tế và các công trình khoa học liên quan đến lĩnh vực nghiên cứu (cid:1)Các phương pháp nghiên cứu cùng lĩnh vực đã và đang thực hiện (cid:1)Thu thập tri thức chuyên môn từ chuyên gia: mẫu câu hỏi tham khảo, phỏng vấn và các đánh giá chuyên môn chuyên sâu

Hai V Pham hai@spice.ci.ritsumei.ac.jp 2

Hai V Pham hai@spice.ci.ritsumei.ac.jp 3

1 9/10/2013

(cid:1) Case Study

Context Matching Algorithm in Searching (cid:1) Context Matching Algorithm in Searching Context Matching Algorithm in Searching Context Matching Algorithm in Searching Alternatives under Uncertain Environments Alternatives under Uncertain Environments Alternatives under Uncertain Environments Alternatives under Uncertain Environments Aware Systems for Intelligent Context----Aware Systems for Intelligent Context Aware Systems Aware Systems for Intelligent Context for Intelligent Context

(cid:1) Thảo luận các nhóm vềdự án môn học với tiến độ giữa học kỳ bao gồm các phần như sau: ◦ 1. Mục đích ◦ 2. Phạm vi ◦ 3. Các sự kiện, ngữ cảnh và cách biểu diễn tri thức ◦ 4. Cộng cơ suy diễn, các luật và diễn giải của luật ◦ 5. Sơ đồ kiến trúc hệ CSTT / hệ chuyên gia ◦ 6. Thiết kế giao diện, giao diện tổng thể và đặc tả chi tiết ◦ 7. Cài đặt chương trình và lựa chọn công cụ lập trình ◦ 8. Kiểm tra và đánh giá ◦ 9. Viết báo cáo tổng kết ◦ 10. Bảo vệ BTL- dự án môn học

(cid:1)

Hai V Pham hai@spice.ci.ritsumei.ac.jp 4

Context Matching Algorithm in Searching Alternatives Context Matching Algorithm in Searching Alternatives Context Matching Algorithm in Searching Alternatives Context Matching Algorithm in Searching Alternatives Aware Systems under Uncertain Environments for Intelligent Context----Aware Systems under Uncertain Environments for Intelligent Context Aware Systems Aware Systems under Uncertain Environments for Intelligent Context under Uncertain Environments for Intelligent Context

Hai V. Pham (Ritsumeikan University) Philip Moore (Birmingham City University

Hai V Pham hai@spice.ci.ritsumei.ac.jp 5

6 Hai V Pham hai@spice.ci.ritsumei.ac.jp

2 9/10/2013

•Research Backgrounds •Research Problem •Context Matching Algorithm •Soft computing integrated with Context Matching Algorithm •Research Discussion •Future works

(cid:1) Context is any information which is used to characterize the situation of entity (objects, activities, preferences,..etc)

(cid:1) Context-awareness means to use context

systems

aim to provide computing information and

information (cid:1) Context-aware searching / communication.

Hai V Pham hai@spice.ci.ritsumei.ac.jp 7

(cid:1) Objects = alternatives has variety of attributes (cid:1) Ex. searching languages

Normal Situation

Natural Situation

◦ something going on ◦ a continuing natural ◦ a series of actions ◦ projecting part of an organism

Action Situation

Biology Situation

(cid:1) Other Ex. Tourism, Business, E-commer,

Translation ..etc

Hai V Pham hai@spice.ci.ritsumei.ac.jp 8

Hai V Pham hai@spice.ci.ritsumei.ac.jp 9

3 9/10/2013

(cid:1) Searching specific majors in languages (cid:1) Recommender alternatives (cid:1) Decision support (cid:1) Tourism Context –aware App. (cid:1) Intelligent Business App. (cid:1) ..etc

Hai V Pham hai@spice.ci.ritsumei.ac.jp 10

We are explored in the area

Hai V Pham hai@spice.ci.ritsumei.ac.jp 11

Hai V Pham hai@spice.ci.ritsumei.ac.jp 12

4 9/10/2013

Hai V Pham hai@spice.ci.ritsumei.ac.jp 13

Hai V Pham hai@spice.ci.ritsumei.ac.jp 14

Hai V Pham hai@spice.ci.ritsumei.ac.jp 15

5 9/10/2013

Hai V Pham hai@spice.ci.ritsumei.ac.jp 16

Hai V Pham hai@spice.ci.ritsumei.ac.jp 17

Hai V Pham hai@spice.ci.ritsumei.ac.jp 18

6 9/10/2013

{IF –THEN} structure, the {IF} operator implementing the component of the rule.

How Rules are affected to alternatives under uncertain environments?

)})})})}

(cid:1) Fuzzy rules ( Human Common Sense

Reasoning)

(cid:1) Self-Organizing Map is used to cluster

alternatives, matched with searching objects

(cid:1) Neural Network is used to train pattern behavior of historical data and predict matched alternatives and objects

R1R1R1R1 IF {{{{ OR }}}} THEN{{{{}}}} (2) R2R2R2R2 IF {{{{ AND }}}} THEN{{{{}}}} (3) R4R4R4R4 IF {{{{ AND AND ((((NOT THEN{{{{}}}} (5) R3R3R3R3 IF {{{{ AND (((( OR )})})})} THEN{{{{}}}} (4) Hai V Pham hai@spice.ci.ritsumei.ac.jp 19

SOFT COMPUTING MODEL

Searching alternatives’ results in dynamic environments under uncertainty

Searching alternatives’ results in static environments

Hai V Pham hai@spice.ci.ritsumei.ac.jp 20

Hai V Pham hai@spice.ci.ritsumei.ac.jp 21

7 9/10/2013

SOFT COMPUTING MODEL (cid:1) SOFT COMPUTING MODEL SOFT COMPUTING MODEL SOFT COMPUTING MODEL (cid:1) SOM (cid:1) NEURAL NETWORKs (cid:1) FUZZY RULES

(cid:1)

Step 1 Step 1: Evaluate the context match {1, 0} for each individual context property, for example: Step 1 Step 1 Step 2 Step 2: Obtain the pre-defined property weighting (wwww) for each context property in the range Step 2 Step 2 [0.1, 1.0]:

(cid:1)

Step 3 Step 3: Apply the weighting (wwww) to the value as derived from step 2 (note: the wwww is applied Step 3 Step 3 irrespective of the value of eeee. Thus retaining the result for eeee):

(cid:1)

IF eeee(a1a1a1a1) = {1, 0} THEN avavavav= (eeee∗wwww)

(cid:1)

Solution 1: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE FIRST CHOICE) Solution 1: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE FIRST CHOICE) Solution 1: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE FIRST CHOICE) Solution 1: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE FIRST CHOICE)

(cid:1)

Step 4 Step 4: Sum the values derived from the CM process: Step 4 Step 4 Step 5 Step 5: Compute the potential maximum value (mpvmpvmpvmpv) for the context properties {a1, b1, b2, c1, Step 5 Step 5 c2}:

(cid:1)

Solution 2: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE SECOND CHOICE) Solution 2: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE SECOND CHOICE) Solution 2: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE SECOND CHOICE) Solution 2: SOFT COMPUTING MODEL INTEGRATED IN THIS STEP ( THE SECOND CHOICE)

(cid:1)

Step 6: Compute the resultant value (rvrvrvrv) for testing against threshold value (tttt): Step 6 Step 6 Step 6

(cid:1)

Matching Algorithm and Soft Computing model results Hybrid Solution: Context----Matching Algorithm and Soft Computing model results (cid:1) Hybrid Solution: Context Matching Algorithm and Soft Computing model results Matching Algorithm and Soft Computing model results Hybrid Solution: Context Hybrid Solution: Context

Hai V Pham hai@spice.ci.ritsumei.ac.jp 22

(cid:1) Thank you for your attentions!

Hai V Pham hai@spice.ci.ritsumei.ac.jp 23

Hai V Pham hai@spice.ci.ritsumei.ac.jp 24