Presentation is loading. Please wait.

Presentation is loading. Please wait.

Presentation Template KwangSoo Yang Florida Atlantic University College of Engineering & Computer Science.

Similar presentations


Presentation on theme: "Presentation Template KwangSoo Yang Florida Atlantic University College of Engineering & Computer Science."— Presentation transcript:

1 Presentation Template KwangSoo Yang Florida Atlantic University College of Engineering & Computer Science

2 1.What is the problem? 2.Why is it important? 3.Why is it challenging (or hard)? 4.What are the limitations of the related work? What is the novelty? 5.What are the contributions? 6.What is the validation methodology? Six Items of Critical Reading

3 1. What is the problem? 1.What is the problem addressed in the paper? 2.Is it clearly stated? 3.Does the author make the important issues clear?

4 2. Why is it important? 1.Is the problem significant / relevant? 2.Is there any reason to care about the problem (practical applications / real world applications) and paper’s results? 3.Is the problem new?

5 3. Why is it challenging? What is the difficulty of solving the problem?  Computationally complex? (NP completeness?)  Large search space?  Hard to solve?

6 4. What are the limitations of the related work? What is the novelty? 1.Has the problem been solved before? 2.Is this a trivial variation on or extension of previous results? 3.Is the author aware of related and previous work, both recent and old? 4.What are the assumptions and limitations of related work? 5.Which limitation is solved in the paper? 6.Critique an assumption that you believe is unreasonable. 7.What is the impact of removing this unreasonable assumption on the solution proposed by the authors?

7 5. What are the contributions? 1.Is the problem the extension of the related work? Which part is extended? 2.Is the problem new? 3.What are the proposed approaches, algorithms, and ideas? 4.Is there any experimental and analytical evaluation?

8 6. What is the validation methodology? 1.Case study? 2.Statistical hypothesis testing? 3.Analytical evaluation? 4.Computational complexity? 5.Theorem and lemmas, proofs? 6.Experimental evaluation? 7.The strengths and weaknesses of the methodology? 8.Why did authors choose this/these methodology?

9 Why is the problem? Input: A spatio-temporal network (STN) Set of operations Output: Storage and access methods Objective: Minimize disk page access Constraints: Spatio-temporal network is too large for main memory. Preserve temporal attribute information. (lossless storage)

10 Lagrangian Approaches to Storage of Spatio-temporal Network Datasets KwangSoo Yang Florida Atlantic University College of Engineering & Computer Science

11 Why is it important? Dynamic social networkAirline flight schedule Power distributionSpatio-temporal road network There are many network applications - Peak load - Cascading blackout - Price - Routes - Schedules - Emerging trends - Evolving network - Propagation models - Dynamic migration - Routing planning - Shortest path

12 U.P.S. Embraces High-Tech Delivery Methods (July 12, 2007) By Claudia H. Deutsch “The research at U.P.S. is paying off. Last year, it cut 28 million miles from truck routes— in good part by mapping routes that minimize left turns.” Societal economic effect Many companies incorporate time-dependent traffic data into network dataset. Speed Profile Why is it important?

13 Time dependent routing planning Minnesota – Austin (TX) Flight Schedule - Average Delay - Flight time Total flight time varies with the start time of day Why is it important?

14 Why is it challenging? A static network contains a large amount of information, temporal information further increases the data size Increased difficulty of rapid data retrieval Nodes Edges Time points: 10,080 (60 mins X 24 hours X 7 days) 10 billion data entries !! : 1,000,000 : 3,000,000

15 Why is it hard? The problem is NP hard problem. The NP hardness of Lagrangian Approaches to Storage of Spatio-temporal Network Datasets follows from a well-known result about the NP-hardness of the balanced min-cut partitioning problem.

16 Basic Concept (Storage Model) Spatio-temporal network representation Time expanded graph Snapshot model

17 GetASuccessor(A,C,1) = C3 : A1  C3 LagGetSuccessors(A,1) = B2, C3 : A1  B2, A1  C3 LagGetPredecessors(D,4) = B2, C3 : B2  D4, C3  D4 LagGetAPredecessor(C,A,3) = A1, A2 A1  C3, A2  C3 1.GetASuccessor() 2.GetSuccessors() 3.GetPredecessors() 4.GetAPredecessor() Parameter : (source, sink, time) Parameter : (source, time) Parameter : (sink, time) Parameter : (sink, source, time) Basic Concept (Access Methods)

18 Related work ST Storage Model Orthogonal Approach Longitudinal Snapshot Lagrangian Approach

19 A  C  D : 2 disk IO Snapshot Approach

20 A  C  D : 2 disk IO Longitudinal Approach

21 Lagrangian Approach A  C  D : 1 disk IO

22 What is the contribution? Proved that the problem is NP hard Proposed LCP partition approach for the problem. Proposed I/O cost model. Experimentally evaluated proposed storage method and I/O cost model against traditional approaches using real dataset.

23 What is the validation method?


Download ppt "Presentation Template KwangSoo Yang Florida Atlantic University College of Engineering & Computer Science."

Similar presentations


Ads by Google