Information Aggregation in an Enterprise - Smart Information Systems: Computational Intelligence for Real-Life Applications - page 116

4.5.4 Web Client
In order to give a complete user experience we have implemented a web client. The
web client is a client-side javascript single-page application implemented using dojo
toolkit. 7 The different result lists, coming fromdifferent broker agents, are aggregated
in the client by sorting the documents according to the normalized score calculated
by the respective broker-agent. Users can also explicitly filter the search results
according to the origin of the network area (main enterprise network, department
network, or local desktop).
Figure 4.8 depicts a screenshot of the graphical user interface of the demo system
that has been deployed in the network of the Berlin administration offices. The
screenshot shows the result list for the search query pia technical documentation
that has been provided by the broker agents of the main enterprise network. On the
left-hand side of the interface, the user can filter the results based on the different
sources. Without logging in, the user will have no access to the department network
and the filter icon for the department network search gives a hint that the user should
first log in.
After successful login, the user is then able to search the department network. By
logging in, the user credential is forwarded from the department network's broker
agent to the search agents in the network area. This credential is used by the search
agents to identify the search results that are relevant for the search query. Together
with the search results from other broker agents the client will then sort these search
results as a single result list. The sorting is based on the normalized score calculated
by each broker agent based on the multiple repositories (Sect. 4.4 ).
4.5.5 Summary
In this chapter we described how our secured distributed enterprise search system is
built and deployed in our pilot project. The network environment in our pilot project
consists not only a main network but also many different department networks. Over-
all we have to consider the following type of network areas with different security
requirements: (1) The main enterprise network is the area where information is avail-
able to all employees as long as these employees are located in the physical network
of the enterprise; (2) the department network is an area where employees must be
authenticated in order to access documents. This authentication is also used to filter
out the relevant documents so that employees only get search results in accordance to
their own access rights; (3) the local desktop is a special area that is only accessible
by the desktop owner. Each of these areas has their own dedicated broker agent. By
using theweb client the employees input their search queries. To receive search results
from department network area the employees must be logged in. Upon receiving a
search request, the contacted broker-agent verifies the user's credential (if required)