By Kevin Davies
February 3, 2009 | Bioalma, the Spanish biomedical IT/text-mining company, has launched a free search tool for the PubMed literature database called novo|seek, which the company claims provides intelligent search functionality to help life scientists guide and refine their searches of the biomedical literature.
The company calls novo|seek “a dynamic information extraction system” for searching biomedical records in repositories, particularly PubMed. Novo|seek indexes the biomeical literature in PubMed and enables researchers to find relevant results efficiently by using external sources of data and contextual term information. The tool provides familiar chronological listings of search results, but a sidebar presents a series of additional related terms based on relevancy, allowing researchers to drill down and refine additional queries.
Bioalma CEO Juan Carlos del Castillo said: “Novo|seek leverages the same Medline data as PubMed, but our search capabilities are second to none and yield better, more relevant results much faster and easier and doesn’t require a lot of training.” novo|seek ranks documents according to their relevance to identify the most interesting information through a recognition of key biomedical concepts. It also ranks authors, letting users know who the main expert is in relation to the query searched.
Search and Deploy
One scientist familiar with novo|seek is Reinhard Schneider, team leader in Data Integration and Knowledge Management at the European Molecular Biology Laboratory in Heidelberg. “I think it’s a very nice tool for researchers, which links the web site with the Deep Web,” said Schneider. “I think it will change the way we work on the web in the future.”
Schneider says novo|seek is a powerful search tool that provides depth of information “typically only found in large text mining machines. With the easy-to-use interface and the ability to filter results in a number of ways the system enables me to drill down to the most relevant information very fast and efficiently.”
Schneider, who has worked with the tool for the past two months, says Bioalma has developed dictionaries of proteins, diseases, chemical names applied to the PubMed abstracts. He calls novo|seek “a tool that allows you to find the most relevant information and put that into context. If you look for diseases, it will give you feedback on what kind of proteins are involved in the disease, what kind of chemicals, and so on. Instead of Googling for hours when you find a new term, you get a very fast overview of what is known about the topic.”
He gives Huntington’s disease as an example. The system returns the standard Medline results, but the sidebar lists a panel of concepts, such as diseases and syndromes, pharmacological substances, proteins, chemical substances, and organisms. “If you click on these terms, you jump from your initial search term to a protein or a gene name. So it’s very easy to get an overview of what is known about the disease,” says Schneider.
Ramón Alonso-Allende, business development director, says Bioalma has been perfecting its information extraction technology for the past five years. “We use statistics, natural language processing and context information. We are able to analyze the literature and extract the knowledge contained [therein] -- we’re not just doing pattern matching.”
Bioalma downloads and indexes 18 million documents in Medline each day. That information is then put into the company’s own database using the open-source Lucene search engine library. “We don’t use NCBI resources to pull data on the fly,” said Alonso-Allende.
General manager Luis Cornide explained that unlike the AKS2 product, which operates in a standard license model, novo|seek operates under a different model. “Our goal is to be there every day with the scientist,” Cornide said. “Novo|seek is designed and conceived to be used by the whole biomedical community, accessed by lots of people to obtain results in Google standard time.”
While Bioalma believes that novo|seek will help introduce its other products to a broader audience, it also hopes to generate revenue by selling online advertizing through Google ads, targeting companies selling reagents or equipment. Bioalma has not held discussions with NCBI about the tool. Alonso-Allende admits that, “Trying to get some of their pie is very challenging. PubMed is a great standard, but it lacks the intelligence that is out there that we can provide with our technology.”
The first release of novo|seek focuses on PubMed, but in time Bioalma plans to integrate additional resources, such as grant information and full-text search.