Presentation is loading. Please wait.

Presentation is loading. Please wait.

Predicting popular areas of a tiled Web map as a strategy for server-side caching Sterling Quinn.

Similar presentations


Presentation on theme: "Predicting popular areas of a tiled Web map as a strategy for server-side caching Sterling Quinn."— Presentation transcript:

1 Predicting popular areas of a tiled Web map as a strategy for server-side caching Sterling Quinn

2 Introduction  This project presents a predictive model for popular areas of a Web map  Model output indicates where server-side cache of map tiles should be created  Selectively caching based on popular map areas can save time and disk space

3 Project objectives  Describe server-side caching of map tiles  Describe the need for selective caching  Present a predictive model for popular areas of the map  Describe ways the model could be used and evaluated

4 Web map optimization and the advent of server-side caching

5 Organizing large maps in manageable “tiles” is not new  Large paper map series are indexed in organized grids  CGIS, a pioneering GIS, used “frames” to organize data (right) From Tomlinson, Calkins, & Marble, 1976, p. 56.

6 Other techniques for organizing maps in tiles or grid systems  Pyramid technique successively generalizes rasters in groups of four cells (right)  Quadtree structures index datasets in a hierarchy of quadrants From De Cola & Montagne, 1993, p. 1394.

7 Server-side caching of map tiles is new  Tiled maps allow users to retrieve just the needed pieces of the map  Cached map tiles give extremely fast performance  Traditional map servers (ArcIMS, WMS) draw the image on the fly  Early static map servers returned the entire map at once

8 Advent of tiled maps and server- side caching  Microsoft Terra Server an early deployment of massive amounts of cached imagery tiles  Google Maps serves cached map tiles with AJAX techniques to create a “seamless” Web mapping experience  Many sites have followed Google’s pattern

9 Caching options

10 Current caching options  Current GIS software allows analysts to create tile caches for their own maps  ESRI’s ArcGIS Server  Mapnik  Microsoft MapCruncher

11 Caching can require enormous resources on the server  Caches covering big areas at large scales can include millions of tiles  Many gigabytes, or even terabytes of storage  Days, weeks, or sometimes months to generate  Many GIS shops lack resources to maintain large caches

12 Selective caching as a strategy for saving resources  Administrator can cache only the areas anticipated to be most visited  Remaining areas can be:  Added to the cache “on-demand” when first user navigates there  Covered with a “Data not available” tile  Left blank

13 Implications of selective caching  Wise because some tiles (ocean, desert) will rarely, if never, be accessed  Requires an admission that some areas are more important than others  Poses challenge of predicting popular areas before the map is released.

14 The need for a predictive model

15 Project presents a predictive model for where to pre-cache tiles  “Which places are most interesting?”  Inputs are datasets readily available to GIS analyst  Output vector features are a template for where to pre-cache tiles

16 Purpose of the model  Help majority of users see a fast Web map while minimizing cache creation time and storage space

17 Not a descriptive model  Descriptive model would show where existing users have already viewed  Microsoft Hotmap a good example of a descriptive tool (right) Microsoft Hotmap Microsoft Hotmap  Descriptive models are useful in deriving predictive models Source: Microsoft Hotmap http://hotmap.msresearch.us

18 Advantages of a predictive model  Doesn’t require the map to be deployed already  Can include fixed and varying geographic phenomena  Has applications far beyond map caching

19 Proposed methods

20 Study area and conditions  Model will predict popular places for a general base map  Study area of California  May create models for thematic maps if time allows

21 Input datasets  Populated / developed areas  Road networks  Coastlines  Points of interest

22 Populated / developed areas  Human Influence Index grid created by the Socioeconomic Data and Applications Center (SEDAC) at Columbia University  Model selects all grid cells over a certain value

23 Road networks  Major roads buffered by a given distance  All roads within national parks, monuments, historical sites, and recreation areas, buffered by a given distance

24 Coastlines  All coastlines buffered by a given distance (wider buffer on inland side)

25 Points of interest  Set of 50 interesting points chosen by model author  Mountain peaks  Theme parks  Sports arenas  Etc.  Represents a flexible layer that could be tailored to local needs

26 Deriving the output  Merge all layers together  Clip to California outline buffered by ½ mile  Remove small holes and polygons  Dissolve into one multipart feature

27 Using the model output  Output is a vector dataset that can be used as a template for creating cached tiles  Compare model output area with total area to understand percent coverage  Compare model output with actual usage over time  Refine if necessary

28 Limitations  Models on a world or country level should account for Internet connectivity  Input datasets have varying collection dates  Input datasets vary in resolution and precision  Maps with many scales might require multiple iterations and variations of the model

29 Questions?


Download ppt "Predicting popular areas of a tiled Web map as a strategy for server-side caching Sterling Quinn."

Similar presentations


Ads by Google