Wednesday 25 October 2009

The International Patent Classification (IPC) is aimed at providing a single scheme to organise and access the world's patent literature. Over time, a number of shortcomings of the IPC system have become apparent, especially as technology is moving rapidly. Consequently, it was decided to reform the IPC. After a short review of the reform, this paper will discuss the impact of the changes on searchers. The possibility of using the reformed IPC to track emerging technologies will be examined through practical examples. At a broader level, the adequacy of such classification systems to describe complex technological information will be discussed.

The European Patent Office is implementing a Machine Translation Programme, with the aim to provide an automated service, capable of translating patent documents from English into the other national languages of the European Patent Organisation member states, and vice versa. The objective is to make the technical content of a patent document sufficiently understandable to a technically qualified person. The basic services of a trial system are fully operational and are internally available to the Office staff. The production system will be made available to the public during 2006. The initial phase is intended to provide the basic infrastructure for the machine translation system and the implementation of the first set of language pairs: English into/from French, German and Spanish. Subsequent phases will be designed to allow the progressive implementation of new language pairs as independent modules of the system, adapted to the pace of co-operation with the corresponding National Offices.

This presentation discusses the market for machine translation technologies, including the questions of Machine versus Human Translation, and free Internet services versus professional solutions. It looks at which applications are best for corporations, and for which users. Case studies are from the the chemical and pharmaceutical domains.
After a study of usage and benefits of translation automation technologies today, the presentation looks at market trends and anticipated new innovations.

In today's safety conscious world there is a strong and increasing emphasis on assessing the possible risk that new therapeutics may pose as early as possible, so that potential failures can be weeded out and resources can be focused on likely winners. Risk assessment is difficult because much of the published safety data is held in separate sources, with inconsistent or non-existent indexing. Even if a source can be identified as potentially relevant, it may take hours of paging through documents to find the required data. One particularly rich source of safety information is the FDA Approval Package for a new drug, and although this can be made available via the Freedom of Information Act, it is not well indexed or searchable, so locating specific facts (side effects, adverse events, etc) is tedious. In this paper we describe work that we have done to these and related relevant sources of safety data (preclinical, clinical and post-market) to make them searchable and accessible so that a wide range of drug safety and risk assessment questions can be answered in a way that can help reduce attrition in the drug development pipeline.

During this presentation we will describe the difference between different data integration layers and the relevance of semantic data integration in the early Life Science Discovery process. The specific examples will show the support of critical elements for the Lead Discovery and Target Discovery process, integrating different sources, scientific objects and information levels.

The key message will be, that the integration challenges of the 90's, which were mostly technology based, are in the final phase of being solved, and that now the industry is facing an even more difficult challenge: the semantic integration, which requires a thorough understanding of the actual data and the business processes which it is used by. Basic access and analysis of data is today's state-of-the-art, but represents only a portion of the user's requirements. To leverage its full capabilities, data and information must be exploited to a further degree.

ChemInfo is a web application developed at Novartis that provides access to a series of structure databases, by structure query or by data search on a few key items. The query is sent to the underlying search systems, results are collected and merged (grouped by structure) and provided as a single hit list. This hit list can be browsed using predefined layouts:

• grouped by structure for a compact overview especially for substructure queries;
• flattened for structure with essential data;
• set of "Substance Sheets" to display all data from the particular data source. Hyperlinks allow to drill-down to further information (eg, to biological data) or to transfer the molfile or Synthesis Sheet to ISIS/Draw.

A search can be restricted to a subset of databases as part of the search constraints or while browsing results. In addition, manual (de)selection is implemented and is preserved when switching from ungrouped to grouped display and back, or when adding or removing data sources. ChemInfo can be connected to any database system that provides either a command-line interface, an Oracle cartridge or a web service and that accepts a structure query in MDL Molfile, Smiles/Smarts or Rosdal format. An ActiveX control is used to transfer queries from or to ISIS/Draw. Display of the query and results is done without any plugin. Data searches focus on category of information, not on data fields in particular databases. Results and the underlying queries are preserved and can be saved permanently by the user.