This document discusses improving temporal search capabilities for spatial data. It proposes using existing metadata to make implicit time information for data layers more explicit. Methods like a "Time Miner" could extract date references from text to enrich metadata. The document also suggests interface enhancements like sorting dates correctly, offering dynamic time bars for filtering, and showing temporal histograms. These changes could significantly increase the value of spatial data by better facilitating exploration and search based on when events occurred over time. The approaches discussed have been implemented in systems like WorldMap and the Billion Object Platform to demonstrate the transformational potential of more fully utilizing time as a dimension.
%in Stilfontein+277-882-255-28 abortion pills for sale in Stilfontein
Making Temporal Search Central in a Spatial Data Infrastructure
1. Making Temporal Search More Central
in a Spatial Data Infrastructure
Benjamin Lewis, Paolo Corti
2017 2nd International Symposium on
Spatiotemporal Computing
Harvard University
2. Thesis
• Time is an underutilized dimension for
improving search in geosystems.
• Straightforward methods exist which can
be transformative for research.
(Better use of the Temporal is just one of several
ways in which SDIs can be modernized.)
5. Systems we have been working on
1. WorldMap – General purpose public
mapping platform. In existence since 2012.
2. The Billion Object Platform (BOP) – Prototype
to lower barriers to access to big streaming
datasets. Recently released.
6. Different systems, common need to
improve search
1. WorldMap (Many Datasets) - Thousands of
data layers with imperfect metadata.
2. BOP (One Big Dataset) - A billion
georeferenced, time stamped tweets.
7. List of enhancements to be discussed
• Time Miner for unstructured text metadata
• Sorting BC / AD dates
• Time bar for date range definition
• Logarithmic time bar increments
• Dynamic temporal histogram
• Ability to zoom on temporal histogram
8. Python TimeMiner
to enrich metadata
• Metadata for map layers is often inconsistent
for time referencing.
• Standard ways of describing date/time such as
ISO 8601 are rarely used.
• Temporal characteristics mentioned as
unstructured text in the title, abstract, and
elsewhere.
9. Initial simple TimeMiner Logic
1. Look for date in the date range section of the
metadata and choose the earlier date. (Date:
from Metadata)
2. If there is no #1 above, look for 4 digit numbers
in title first, then abstract, which are less than or
equal to 2017 (present year) (Date: Detected)
3. If there IS a date in #2 above, check to see
whether there is a CE or AD or BCE or BC after it
and apply math accordingly (Date: Detected)
4. If there IS NO #2 above, look for 1, 2, or 3 digit
numbers with associated CE, AD, BCE, BC, and
apply math accordingly (Date: Detected)
10. Another technique: Historic Periods
Example: Chinese Dynasties
• Xia, Hsia ca. 2100-1600 BCE
• Shang ca. 1600-1050 BCE
• Zhou, Chou ca. 1046-256 BCE
• Qin, Ch'in 221-206 BCE
• Han 206 BCE-220 CE
• Sui 581-618 CE
• Tang, T'ang 618-906
• Song, Sung 960-1279
• Yuan 1279-1368
• Ming 1368-1644
• Qing, Ch'ing 1644-1912
Source: http://afe.easia.columbia.edu/timelines/china_timeline.htm
17. Conclusion
• Most spatial data describes events in time though
often not explicitly.
• When data does have a time component it is
often not easily accessed.
• An opportunity exists to increase the value of
existing data by:
1. Making latent temporal information explicit using
enrichment techniques
2. Implementing UI/backend enhancements on existing
systems
3. Increase research on space/time data exploration
18. More information
WorldMap / HHypermap
• Information
http://gis.harvard.edu/publications/implementing-
open-source-spatiotemporal-search-platform-spatial-data-
infrastructures
• Source code https://github.com/cga-harvard/HHypermap
BOP
• Information http://gis.harvard.edu/services/project-
consultation/project-resume/billion-object-platform-bop
• Source code https://github.com/cga-harvard/hhypermap-
bop
Today I’m going to talk about work we have been doing at CGA to develop Temporal search capabilities for some geospatial data systems.
As with most of the work we do here, its undertaken to solve practical problems, problems we faced developing mapping systems for serving an interdisciplinary user community.
I will present some ways we are addressing this through data enrichment and UI improvements.
This is a work in progress.
As an example, here we have the UI for the main U.S. Govt Geo-Data search tool.
Here is the GeoNode search UI. Buried down here we have the option to key in a start date and an end date.
Different contents require different approaches but still there in much in common between the two approaches.
Being able to search through large holdings and find what one needs for a particular purpose is key in both systems.
In addition to keyword, spatial location is a useful filter because all records have a spatial extent. Space cuts across disciplines nicely.
The attempt to add time. Why time? Everything happens sometime. Cuts across disciplines as with space.
Any dates determined using the TimeMiner should be denoted as “detected”. Those in metadata will be “from metadata”
https://en.wikipedia.org/wiki/ISO_8601
With demarcations from metadata or detected.
Use period name string to define time periods. Can be used with spatial extent to improve accuracy.
This approach could be applied to other situations where reasonably unique strings are associated with defined time periods. As a starting point:
https://en.wikipedia.org/wiki/List_of_time_periods
ca. = circa = "more or less" or "approximately"
BCE = Before the Common Era
B.C. = “before Christ”
A.D. = Anno Domini (Latin : "In the year of (our) Lord")
Ability to sort results in UI
Date detected
Time bar past
Time bar future end
The idea here is to give the human user the tools and visual feedback needed to efficiently explore a large collection.