Scientific RDF databases

stickyraffleΛογισμικό & κατασκευή λογ/κού

4 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

86 εμφανίσεις

Scientific RDF databases

To describe objects in the sematic web, RDF is mostly used. RDF descriptions are based on triples to
describe the data. The triples are composed of a subject, a predicate and another object. By referring
to other objects, which
are possibly described by some
one else, linked data is formed. When
compared to relational databases, RDF databases have several advantages. They are for example
very easy to extend and have a scalable, distributed design. The data in RDF databases is also

self
-
documenting. An example site that uses RDF is “open library”. It describes books in the RDF format.

RDF databases and constraints

How can we start from relational databases and convert them to RDF databases and also keep the
constraints that were on
the relational database. We first need to look at the constraints that are
available in the RDF and then try to map the constraints within the relational model to constraints in
the RDF. RDF has two types of constraints, vocabulary constraints and non
-
voca
bulary constraints.
The const
raints can be expressed in SPAR
QL.

Databases and computer games I

Computer games that use large databases to store their game state, for example MMO games,
require fast updates to provide a good user experience. It is also nec
essary that checkpoints are
created regularly so that the game state

is as accurate as possible when a server crashes and needs
to be restored. There are different ways for updating the game state: naïve snapshot, which
performs better on very large number

of updates; copy on update, which copies only those objects
that have changed; and wait
-
free
-
pingpong, which is best for eg. Less than 1000000 updates per tick.

Databases and computer games II

SGL is the Scalable Gaming Language. It is a scripting languag
e to specify individual unit behavior. It
consists of two parts. The data definition file which contains the schema for a unit. The second part is
the script file which represents a single action for a single unit. At each clock tick we take our current
en
vironment and create a new environment using the actions in the script file.

The script file consists
of SQL together with if then else statements and other statements needed to create a script.

Mapreduce

MapReduce was developed by google to support distri
buted computing on large datasets on
computer clusters. The purpose was to abstract the distributed computing part so the developer can
focus on the development of the algorithm and not on the development of the distributed
computing part. The programming
model consists of two functions: the map function and the reduce
function.

Mapreduce demo

Hadoop is an implementation of mapreduce. It provides an interface to use mapreduce. There is a
command line version of hadoop, but it is easier to use a graphical us
er interface like for example
HUE (Hadoop User Experience). In HUE it is easy to specify SQL
-
like queries to query the data in the
Hadoop Distributed Filesystem (HDFS). There is also an easy interface to view the tasks on the
jobtracker, the tasktracker an
d the namenode.

Web link structure and its usage for sponsored search markets

Original pagerank: pages with many visits/clicks are ranked higher. Problem with this is that it is easy
to manipulate. A new pagerank algorithm, developed in 1998, checks how ma
ny references are made
to a page from other pages. Popular pages have many references to them so they are ranked higher.
Another algorithm is HITS (Hyperlink
-
Induced Topic Search). SEO or Search Engine Optimization is the
process of
improving

the visibility of a website or a webpage in search engines via un
-
paid search
results.

There are two types of SEO, black hat and white hat. White hat is the ethical way, while black
hat uses non
-
ethical techniques to improve the pagerank.

Network dynamics
: population models


information cascades

The whole world is connected: we are only six connections apart from any other person in the world.
We are affected by the decisions in our environment. Experiments show that people tend to follow
the meaning of o
thers rather than to rely on their own information when a
lot of people think
differently (marble experiment).