Cold War Kuhn and industrial science
May 5, 2010 at 12:59 pm | Posted in book | Leave a commentI read Kuhn early on in my PhD work, and was much impressed, as many before me, by Kuhn’s portrayl of a science governed by paradigms and puzzle-solving. His writings have served a useful purpose thus far in my thinking, which is to undermine any traditional and mythological preconceptions I may still hold about science.
What I have taken from Kuhn, and to a lesser extent from Merton and Polanyi, is that there are other important elements in the story which is the way science is done, and many of these elements are quite unscientific.
However, I feel my philosophical preferences fall firmly on the side contra-sociological models for science and I believe rationality, realism and critical argument do play a part in what science *is*. Scientific knowledge is special and different from lay-knowledge – we can test theories, we can challenge our models of the world and thus aspire to a bit more truth tomorrow to the truth we have today.
Fuller’s book casts Kuhn in a Cold War context, and convinces me at least that a lot of what Kuhn was describing in his historicist re-envisioning of scientific enterprise was coloured by his pre-dilections for country and conservatism. No liberal was Kuhn, and in the Cold War climate his painting of science as ruled by paradigm and a theoretical conservatism fitted with an American way of thinking about the world, with loyalty to codes and just causes.
The Popper bits are a bit superfluous. In a simple way though, Popper’s critical liberalism does contrast with Kuhn’s conservatism.
And it is interesting to think that Kuhn’s view on science, and the popularity of social studies of science today, stifle criticism in science and serve to propagate an industrialised science.
Fuller, S., Kuhn vs. Popper : the struggle for the soul of science. (Columbia University Press, New York, 2004).
For and against methods in ontologies
May 5, 2010 at 10:22 am | Posted in book | Leave a commentMy interest in ontologies and specifically in ontologies for biology have compelled me to extend my reading into the philosophy of science. If an ontology is a representation of some knowledge and how the named elements in this knowledge relate to one another, then what are the criteria for admitting new knowledge into this structure?
This is an old problem in the philosphy of science and can be variously phrased as the demarcation problem, the way to separate science from pseudoscience, or put simply – how scientists choose between good theories and bad theories?
Science clearly changes over time, and old theories morph or become substitued by new and, in some sense, better theories. Lakatos’s series of lectures published in the book ‘For and against method’ (reference below) are a superb introduction to this fascinating area, and his style is frequently hilarious:
In 1492, however, Columbus had discovered America, and that led to some trouble because in 1949 an American named Alonzo Church reviewed Ayer’s book in the Journal of Symbolic Logic. The review is actually twelve lines long.
It really makes me wish I’d been born a few decades earlier so I could have listened to lectures at the LSE with people like Popper and Lakatos. The problems which preoccupied these men have seemingly been sidelined to the sociological strands in the study of science.
However I believe the development and application of ontologies in the sciences force us to confront old questions of the kind that interested people like Lakatos. How or who gets to decide what goes into an ontology?
What is the method for deciding how an ontology changes?
Lakatos, I., Feyerabend, P., & Motterlini, M., For and against method : including Lakatos’s lectures on scientific method and the Lakatos-Feyerabend correspondence. (University of Chicago Press, Chicago, 1999).
Notes on Foskett's 'The subject approach to information'
August 3, 2009 at 3:05 pm | Posted in book | Leave a commentTags: classification
A.C. Foskett (1983). The subject approach to information, 4th edition.
Notes on a relatively old book, ‘The subject approach to information’, which I think is still pertinent in the light of efforts by the biology community to produce subject lists / vocabularies / ontologies to aid in annotating gene product entries in databases. In particular, Gene Ontology is a good example of one such controlled vocabulary used to classify or index entries in all sorts of online databases.
- Gene Ontology is equivalent to what Foskett refers to as an ‘indexing system’
- Gene Ontology is an indexing system for gene sequence databases, gene product databases, and is also used as an indexing system to access empirical data or as a tool in processing natural language in the scientific literature.
- pp.2-3 : Mention of growth of knowledge in the 20th century (a la Popper) with the challenge being for individual’s to incorporate new knowledge into their personal knowledge structures
- pp.4 : “…a set of records (surrogates) of the content of a library”; gene names and associated records are surrogates for sequences in the genomic library
- The need to achieve access to particular items via different approaches has not changed since the 4th edition was published eg, in moving from the specific (‘cytochrome C’) to the general (‘protein kinase activity’)
- Index terms strive to mark records as ‘relevant’ to the user. How can we measure GO term relevance to the thinking biologist?
- GO terms are used both to aid retrieval / organisation, and to support logical inferencing by computers
- pp.16 : Retrieval is needed to keep up-to-date with a changing (very relevant to biology); also science terminology is referred to as ‘hard’ and science documents as ‘self-indexing’. Doubt many bioinformaticians would agree with this today.
- What is ‘interesting’ to a biologist?
- What does a biologist need information for?
- Error – recall – precision – specificity – exhaustivity : how does GO term annotation influence these?
- pp.32-33 : “The heading is the subject description which determines where in the sequence we shall find any give entry”
- Is the Gene Ontology alphabetical or systematic?
- A description provides information on the factors that serve to identify an entry / document / gene product; can be an accession number
- subject entry = heading + description
- index entry = heading only (leading to entries)
- How is a GO term assigned? Is it based on sequence homology? Natural language processing? Differential gene regulation?
- pp.69 : semantic factoring
- pp.72-73 : semantic relationships = equivalence, hierarchical and affinitive / associative
- How does the history of biology, especially taxonomy, determine the hierarchical design of indexing systems like the Gene Ontology?
- pp.76 : orphan terms in the Gene Ontology would be those terms with no associated gene products, eg, the biological process GO term ‘flight‘
Some criticisms of Popper's World 3
June 26, 2009 at 10:49 am | Posted in book | Leave a commentTags: epistemology, knowledge, popper
Notturno, Mark Amadeus (2000). The meaning of World 3, or why Wittgenstein walked out. In Science and the open society : the future of Karl Popper’s philosophy. Central European University Press.
- Of the purpose of Popper’s 3 Worlds epistemology:
“It is, in fact, proposed to solve two problems at once: the problem of objective knowledge on the one hand, and the problem of the relationship between the body and the mind on the other.” (pp. 144)
- Philosophers think that Popper’s World reifies knowledge and contradicts his own theory that scientific knowledge is fallible. However Popper’s World 3 is quite different to the Platonic realm of ideas in that it evolves – it is objective / autonomous AND fallible.
- Notturno argues that most criticisms of Popper’s epistemology implicitly accept empiricist dogma.
“How is objective knowledge possible – given that the only things that exist are those that can be known through the senses.“(pp. 146)
- If one accepts empiricist dogma, then one simply cannot accept Popper’s epistemology because it includes immaterial entities. World 3 cannot exist because we cannot prove it exists by inspect it through our senses.
- The 3 Worlds contravenes the parsimony argument, that theories should be simple. A good point Notturno makes is that Popper did not start with World 3 – he was led to incorporate World 3 into his epistemology as a solution to the problem of objective knowledge. Solutions without World 3 are weak, or implausible.
- Another problem: philosophers are uncomfortable with Popper’s epistemology because it argues that the immaterial can affect the material. It extends the problem of Cartesian dualism and causality – or how the body and mind interact – to create a supplementary set of interactions between World 3 and World 2.
- Popper’s 3 Worlds is a solution to a philosophical problem, but Wittgenstein would say that there are no philosophical problems, so the epistemology is garbage.
Are Premiership footballers more important than scientists?
June 22, 2009 at 9:29 am | Posted in book, web | Leave a commentTags: knowledge, wisdom
Reading an interesting book by Nick / Nicholas Maxwell called ‘From knowledge to wisdom‘.
Maxwell argues for a shift in academic inquiry from the acquisition of knowledge to the acquisition of wisdom, wisdom being “…the capacity to realize what is of value in life, for oneself and others.”
For example science has created great technological developments in the last century, yet still the human species is plagued by war, hunger, poverty, inequality, disease and social malaise. If all academic inquiry were grounded in an aim-orientated empiricism, in seeking to realize what is of value in life – to be happy, to love, to co-operate, to be fair – more might be done to solve these seemingly intractable problems.
The appeal of ‘From knowledge to wisdom’ for me is in the challenge for scientists to answer the question: what is the purpose of research? If it is make the world a better place, the present system of academic inquiry is woefully designed to achieve this. If it is to acquire knowledge for knowledge’s sake, that it might be applied to solve hunger, or inequality, or deficiencies in the well-being of our species, then where is the evidence that science is indeed leading us to realize satisfaction in life?
Lousy jobs, alienation, disempowerment, ill-health, financial inequalities, low culture: we might all be vaccinated and have flatscreen televisions, but does the man on the street believe a scientist is more or less important to his life than a Premiership footballer?
For more information on Maxwell’s thesis, see www.knowledgetowisdom.org
Scientific practices or scientific products?
June 17, 2009 at 3:03 pm | Posted in book | Leave a commentTags: information practices, science
Wouters, P., Vann, K., Scharnhorst, A., Ratto, M., Hellsten, I., Fry, J., and Beaulieu, A. (2008). Messy shapes of knowledge – STS explores informatization, new media, and academic work. In Hackett, E. J., Amsterdamska, O., Lynch, M., and Wajcman, J., editors, The Handbook of Science and Technology Studies, pages 319-352. MIT Press.
I know that one cannot entirely separate the study of the practice of producing scientific knowledge from the study of the scientific product itself. Yet a conflation persists in the information science and science studies literature: the conflation of scientific practices with scientific products.
The error is even made explicit in document titles. ‘Messy shapes of knowledge’ suggests one is interested in the nature of knowledge itself, in its characteristics, tendencies, colourings.
Instead, the aforementioned title precedes an essay on the practice of scientific knowledge generation. How are new technologies changing the way science is done? What are the implications of new technologies – such as the Internet – for science as work and for scientific knowledge as a labour product?
The scientist is the object of study. He/she is typed, inspected, pondered. The knowledge product is so obvious as to not warrant attention. If scientific knowledge is to be considered, it is only in:
“[...] the way empirical materials and facts are combined to produce a plausible story or vision of the future [...].”
In the study of science and technology, knowledge as an objective entity is often denigrated. To appeal to an objective knowledge is to ‘reify’ but a version of all we might know. Every individual’s knowledge is primary and thus, no knowledge is primary. A shared scientific consensus is oppressive, stifling, institutional.
I resist the the demotion of objective scientific knowledge. By objective I mean in Popper’s World 3 sense, as a knowledge external to the mind of Man. I put scientific knowledge on a plinth. I am interested in products over processes.
Jazzing up biomedical text mining with ontologies
April 3, 2009 at 8:27 am | Posted in book | Leave a commentTags: semantic web, text mining
Witte, R., Kappler, T. & Baker, C.J.O. (2007). Ontology design for biomedical text mining. In Baker, Christopher J O, Cheung, Hei-Hoi (Ed.), Semantic web : revolutionizing knowledge discovery in the life sciences (pp. 281-313). Springer.
See this book on the Publisher’s website
- Text mining is important to biology because it is a cheap alternative to labour-intensive, manual curation of the rapidly expanding literature
- Text mining is natural language processing of non-structured texts to extract salient facts
- Do you need an ontology to do text mining? No.
- Formal ontologies facilitate data integration from different sources
- Ontologies can also be used to pose questions
- Blunt information retrieval, or applications using simpler taxonomies and thesauri can also be used to integrate data sources and ask questions
- Ontologies are ‘richer’ knowledge representations (to use a highly subjective term), though are they demonstrably better for text mining? How can we test how useful an ontology is?
- The authors describe adding ontology support to an information system that annotates protein sequences. The annotations are derived from the literature, and contain information about the functional effects of different mutations.
- It would be interesting to see if the system performed better with ontology support, and if so, how much better.
- Many ontologies are built on existing, manually-curated taxonomies or classifications
“So what precisely are the benefits again? [...] In short, exporting NLP into an OWL-DL ontology (ontology population) allows for standardised data exchange, which in particular includes reasoning tools that can be used to query the ontology [...] Using an ontology during NLP analysis allows one to consolidate the various resources, stored in different representational formats, into a single datastructure, thereby ensuring semantic integrity between the various analysis steps.” (pp.309)
In short, an ontology allows one to ask questions and harmonise different data sources.
OWL for the novice, ontologies for who?
April 3, 2009 at 7:47 am | Posted in book | Leave a commentTags: ontology, semantic web
Pan, J. Z. (2007). OWL for the novice : a logical perspective. In Baker, Christopher J O, Cheung, Hei-Hoi (Ed.), Semantic web : revolutionizing knowledge discovery in the life sciences (pp. 159-182). Springer.
See this book on the publisher’s website
Do we need ontologies and corresponding languages like OWL? Pan explains how ontologies can be used to:
- Justify classes in a taxonomy, and classify new instances
- Detect errors in database entities and queries
- Generate UML Class diagrams in software engineering
The list is not very long. Dp ontology languages like OWL exist to help people, or to help machines help people, or to help machines help themselves, and can machines ever actually help people, and would it not be preferable for people to help people?
- OWL stands for Web Ontology Language (why not WOL?)
- OWL is/was produced by the W3C
- OWL is designed to enable people to analyze and manage information in a *meaningful* way
- To achieve this goal, web content needs semantic markup
- Until now, the Web has mainly be marked-up with syntactic tags
- OWL has developed from the Resource Description Framework (RDF)
- RDF and OWL are first-cousins. RDF is a standard for annotations of the form [subject property object]. RDF and OWL differ in their semantic models, and hence their expressive power.
- OWL has three sub-languages: OWL Lite, OWL DL and OWL Full
- OWL Lite and Owl DL are very expressive description logics. They have well-defined semantics and formal properties, especially decidability and complexity of reasoning
- OWL Full is undecidable
- OWL languages have a syntax (layout) and semantics (reasoning)
- There are special querying languages for ontologies
What semantic web tools can do for translational medicine
March 31, 2009 at 11:37 am | Posted in book | Leave a commentTags: biology, information management, knowledge, semantic web
Kashyap, V. & Hongsermeier, T. (2007). Can semantic web technologies enable translational medicine?. In Baker, Christopher J O, Cheung, Hei-Hoi (Ed.), Semantic web : revolutionizing knowledge discovery in the life sciences (pp. 249-279). Springer.
See this book on the Publisher’s website
- Translational medicine is a popular buzz-term in the life science right now, and means the transfer of bench-side discoveries into clinical applications
- The translational medicine problem is seen as one of communication – enabling communication between basic scientists and clinicians
- Information flows occur between : patient clinical data, clinical trials, tissue banks, genetics support services, bench R&D, genomic databases
- Components in the information architecture for translational medicine include : portals, applications, data repositories, decicision support services, knowledge asset management, data support
- Semantic tools are seen as a solution to the problem of integrating these different information sources and components
- A translational medicine ontology would integrate concepts from the clinical field (patient suffers_from disease) and the basic sciece domain (diagnostic test identifies_mutation gene)
- RDF wrappers will translate elements in Laboratory Information Managment Systems (LIMS) across into Electronic Medical Records (EMR)
- The hope is that services integrated using semantic tools will significantly aid clinical decisions, and facilitate knowledge management in the clinical setting
Problems with rummaging the biomedical literature for knowledge
March 30, 2009 at 9:52 am | Posted in book | Leave a commentTags: biology, semantic web, text mining
Hirschman, L., Hayes, W. S. & Valencia, A. (2007). Knowledge acquisition from the biomedical literature. In Baker, Christopher J O, Cheung, Hei-Hoi (Ed.), Semantic web : revolutionizing knowledge discovery in the life sciences (pp. 53-81). Springer.
See this book on the Publisher’s website
- Biomedical literature is viewed as a repository for biomedical knowledge. It is assumed the repository contents are true.
- The biomedical literature contains evidence, procedures and reasoning. We might usefully extract these components.
- The semantic web for the life sciences is to aid: (1) Navigation and, (2) Calculation
- The biomedical literature is big, growing, dispersed and goes has a ‘use by’ data
- Biomedical scientists want to ask questions the literature questions, including open-ended, closed-form and specific
- Currency is important – when is the information from? Is it out of date?
- The semantic web will impact costs, such as the cost of curating or not curating data, or the cost of not exploiting the existing literature
- Quality is also important – does the information meet the scientist’s exacting standards?
Some semantic web tools for the life sciences mentioned include:
- Automatic annotation of public sequence databases (including text mining to capture context information such as experimental conditions)
- Model organism databases (in other words, curated exemplar genomes, with mapped entities and their functions)
- Drug discovery (by natural language processing of the literature to infer new drug-target relations)
- High-throughput data interpretation (such as annotating gene sets from microarray experiments)
Technical limitations, or information problems, mentioned include :
- Entity identification
- Designing tools for new tasks
- Creating curation tools to speed up manual curation
- Ontology maintenance, mapping and versioning
- Full text access
- Systems evaluation
- Database interoperability
- Ontology versioning
- Currency in all semantic web tools
Blog at WordPress.com. | Theme: Pool by Borja Fernandez.
Entries and comments feeds.
