Most Informative Attribute

Artist s IDT algorithm is an adapted version of a Quinlan-like algorithm described in (Bratko, 1990). The result is represented in the form of a Decision Tree (DT), where internal nodes are labelled with attributes, and branches are labelled with attribute values (note, however, that the DT is not the same as the ASS representation discussed earlier). The leaves of the tree are labelled with sound classes. To classify a sound event, a path in the tree is traversed, starting at the root node and ending at a leaf. The IDT algorithm (refer to Appendix 3) proceeds by searching at each non-terminal node for the attribute whose values provide the best discrimination among the other attributes, that is, the Most Informative Attribute (MIA) the formula for the selection of the MIA is explained in Miranda, 1994. 

Each time a new MIA (Most Informative Attribute) is selected, only those attributes that have... 

Personal computers have introduced new ways to handle pdvate biblographic and text files. The most important factors to consider to achieve satisfactory results in building a bibhographic or text database are the type of information to be stored and the needs of the user. Types of information include correspondence, research results and documentation, meeting notes, and bibliographic references. Needs of the user to be considered should include the potential number of users of the database, restrictions for the access and display of the information because of privacy or proprietary reasons, and the retrieval mechanisms (eg, by keyword, authority list, controUed vocabulary, author, tide, date, or other document or information attributes). In addition, criteria for selecting and encoding information for the database need to be established. 

In chemical kinetic studies the most relevant attributes are the counts of the various species present and the numbers of transitions of various types that occur during each iteration. For example, in a study of three types of reacting ingredients, A, B, and C, the numbers of each species will change with time, and this variation can reveal important information about the kinetics of the reactions involved. Also informative will be the numbers of transitions, say, from A B and A C, that take place in each iteration. 

As pharmaceuticals and their metabolites find their way to the environment primarily via the discharge of raw and treated sewage from residential users or medical facilities, most information on their occurrence and fate is related to WWTPs and the receiving waters. Much less is known about their presence in DWTPs and finished (tap) water. This can be attributed not only to the performances of the analytical technologies to detect a diverse range of low levels... 

When the reporting rate indicates a frequency of occurrence apparently lying in the range 1/1000 to 1/10,000, an accurate overall idea of incidence and attributable risk is desirable in addition to qualitatively defining risk factors. It is not easy to generalize about the type of study which would be most informative, but the use of computerized data bases... 

In the empirical valence bond (EVB) model [304, 349, 370] a fairly small number of VB functions is used to fit a VB model of a chemical reaction path the parameterisation of these functions is carried out to reproduce experimental or ab initio MO data. The simple EVB Hamiltonian thus calibrated for a model reaction in solution can subsequently be used in the description of the enzyme-ligand complex. One of the most ingenious attributes of the EVB model is that the reduction of the number of VB resonance structures included in the model does not introduce serious errors, as would happen in an ab initio VB formulation, due to the parameterisation of the VB framework which ensures the reproduction of the experimental or other information used. This computationally efficient approach has been extensively used with remarkable success [305, 306, 371, 379] A similar method presented by Kim and Hymes [380] considers a non-equilibrium coupling between the solute and the solvent, the latter being treated as a dielectric continuum. 

Skin reactions are often reported with NSAIDs, but the true incidences with individual NSAIDs are unknown. There are very few specific epidemiological studies, and most information comes from single case reports and data from national spontaneous reporting systems. A major study on nearly 20 000 patients showed that 0.3% of 9118 patients taking analgesics and NSAIDs developed skin reactions that could be attributed to these drugs (164). 

Before moving on we briefly examine the key characteristics of the different experiments and consider why these might be of interest to the research chemist. Table 5.5 summarises the most significant attributes, some of which have already been introduced whilst others are expanded in the sections that follow. Whilst the TOCSY experiment is not strictly a member of the COSY family, its information content is so closely related to that of COSY it has also been included in the table. 

The lanthanide patterns, normalised to the chondritic abundances, are shown in figs. 8 and 9 and the data are listed in tables 10 and 11. Several interesting features emerge. The most informative is the depletion in europium which is discussed in the next section. Secondly, the abundance patterns are broadly parallel to one another, but with some differences in detail. The generally parallel nature of the total patterns (excluding Eu) over more than an order of magnitude is attributed to two causes. 

Hypermedia permits the user to determine the sequence in which to acess information (browsing), to add or amend the information in order to make it more personally meaningful (collaboration), or to build and structure their own knowledge base. Like most information systems, interaction is one of the most important attributes. Hypertext information systems permit more dynamic user control of the information in the knowledge base. 

The intersection of the microscopic scale with information presents a vast literature. To list a sampling most helpful to the author, one begins with the information theory and statistical thermodynamics work of Jaynes [4], and the later text by Baierlein on atoms and information [5]. At a less advanced but still highly illuminating level are books by Morowitz [6,7]. Information casts a wide net in chemistry. Levine and coworkers have long championed information theory applied to molecular processes such as relaxation and internal energy redistribution [8,9]. Biopolymers plus information yield the field of bioinformatics. Recommended is the text by Tramontano for the landmaik questions posed [10]. The research of Schneider has addressed in depth the information attributes of biopolymers [11,12]. 

It was reahzed quite some decades ago that the amount of information accumulated by chemists can, in the long run, be made accessible to the scientific community only in electronic form in other words, it has to be stored in databases. This new field, which deals with the storage, the manipulation, and the processing of chemical information, was emerging without a proper name. In most cases, the scientists active in the field said they were working in "Chemical Information . However, as this term did not make a distinction between librarianship and the development of computer methods, some scientists said they were working in "Computer Chemistry to stress the importance they attributed to the use of the computer for processing chemical information. However, the latter term could easily be confused with Computational Chemistry, which is perceived by others to be more limited to theoretical quantum mechanical calculations. 

For purposes of dosage, the specific activity of an enzyme is usually expressed as International Units (lU) rather than in terms of weight. However, unit measurements do not provide information on the absolute purity of a given product. Moreover, purity is not as critical an attribute for oral enzymes, as opposed to those adrninistered parenteraHy, inasmuch as the gastrointestinal tract is capable of disposing of most inert contaminants. 

The effects attributed to air pollutants range from mild eye irritation to mortality. In most cases, the effect is to aggravate preexisting diseases or to degrade the health status, making persons more susceptible to infection or development of a chronic respiratory disease. Some of the effects associated with specific pollutants are listed in Table 7-2. Further information is available in the U.S. Environmental Protection Agency criteria documents summarized in Chapter 22. 

---