Most data available on the World Wide Web (hereinafter Web) can be obtained only by means that are friendly for Web users, but not useful for automatic and mechanical processing by software applications. Nowadays, many Web sites offer ad hoc query interfaces with forms that return the data required in lists comprising semi-structured responses encoded in HTML documents. This part of the Web - accessed through different types of forms and/or interfaces that return data automatically obtained from internal databases - is normally called “Hidden Web”).
This “Hidden Web” is by no means a small part of the whole WWW and contains a huge amount of data which, in many cases, are of great quality and interest to users. Web sites like e-shops (that provide their catalogs in this way) and search engines for data of a scientific, health, patenting or financial nature are good examples of this. It is also often the case that these Web sites are private access (i.e. a user/password is required to access them), have an advanced query interface (allowing data searches in respect of different subject matters) and/or return results in the form of lists of items encoded in HTML with links to related pages that contain more data on each item (e.g. generally e-shops return a list of results, but with the option for the user to “click” on the title to access another page with commentaries on the product, photos, related products, etc.).
In addition to the problem of accessing these sources with “hidden” data, applications that want to use these data are also frequently faced with the problem of results being returned in HTML, which is a tag language defined for visual display by users that never publishes metadata of any type on the structure and/or semantics of the results generated. Neither does it structurally differentiate navigation elements (menus), graphic panels and data useful to the user. The problem of extracting the relevant data contained in HTML pages thus also arises.
Example: Look at this example of an Internet bookshop with a search form as shown in the figure below. The form obliges users to specify a value for the attribute “TITLE” and gives them the option of entering a value for the attribute “AUTHOR” and for the attribute “FORMAT” (restricting a group of values).
The bookshop returns a result list with data on TITLE, AUTHOR, FORMAT, PUBLISHER and PRICE.
This case summarizes the difficulties an application faces when attempting to extract structured data from Web environments: accessing Web sources, navigating through transactional environments, option selecting and, finally, extracting data from semi-structured data.
Denodo ITPilot is the Denodo Technologies solution for easy access to and structuring of datasets on the Web; this process involves constructing an abstraction from the specific Web source called wrapper that isolates the client applications from the intrinsic characteristics of this site (access protocol, native data structure, etc.). ITPilot provides a distributed and scalable environment for generating, executing and verifying wrappers.
This manual presents Denodo ITPilot and provides instructions for correct installation, recommendations on the different types of architecture it supports, as well as a guide to the execution and verification environment:
- The components of ITPilot are introduced in this same section
- The next section will provide an overview of the recommended architectures.
- The section Installation and Initial Configuration gives a detailed description of the installation process for each of the components
- The section Introduction does the same job with the server startup procedure.
- The section Web Administration Tool explains the ITPilot Web Administration Tool.
- The section Verification Server Graphical Configuration and Monitoring Tool deals with the ITPilot Verification Server Graphical Configuration and Monitoring Tool.
- The section Wrapper Execution Trace Logs gives instructions on how to set up logging for the trace events generated during wrapper execution.