PENERAPAN METODE CONSPECTUS UNTUK MENGUKUR INTENSITAS KOLEKSI MONOGRAF DI BADAN PERPUSTAKAAN DAERAH PROPINSI DAERAH ISTIMEWA YOGYAKARTA PERIODE PENGADAAN TAHUN 2003

"The Application of Conspectus Method to Measure The Intensity of Acquisition for Monograph Collection at Badan Perpustakaan Daerah Propinsi Daerah Istimewa Yogyakarta period 2003" is the main purpose of this study. The writer formulates three main problems, that is how to apply conspectus method, how to know the intensity of monograph collection, and how to understand strength and weakness of the monograph collection at Badan Perpustakaan Daerah Propinsi Daerah Istimewa Yogyakarta. The type of this study is a quantitave description research. The main classification of monograph collection 000 — 900 period 2003 is as a subject and Badan Perpustakaan Daerah Propinsi DIY is as an object. The sampling was taken 50% from the total population. The variable researched is the distribution of classification, strength and weaknesses of the collection, the chronological and language analysis. The method of searching are field research, observation, interview guide and inside evaluator. Data analysis is the process to describe the collected data during this research. Then the result is applied toward conspectus method based on 1 — 5 and language code. Conspectus method is to value library collection based on subject area and to describe the srength and weaknesses of the collection, and to know collection level at library. The result of the study indicates that the Badan Perpustakaan Daerah Propinsi DIY has not applied yet the conspectus method to value strength and weakness of the monograph collection it only uses statistics of acquisition for monograph collection. The intensity of acquisition for the monograph collection is high, it is on classification 300 (social science) and classification 000 (general literature) the result of the inside evaluator is lb to 2a. la (Minimal Level) means the library has a limited core literatur, however it has amount of literature written by main writers. 2a (basic Information Level) means that core material is available to define a subject. Chronological inclusion of monograph collection is relevant, because the needs modern information mainly is 10%. Material literature published in 2003 is 28%. Language analysis of monograph collection is Y, it means the literature material is dominated by one language, except English. Key word : Conspectus, Intensity of Collection Acquisition

Comparing the performance of different stream classification systems using aquatic macroinvertebrates

AIM: We evaluated five stream classification systems observing: 1) differences in richness, abundance and macroinvertebrates communities among stream classes within classification systems; and 2) whether classification systems present better performance using macroinvertebrates. Additionally, we evaluated the effects of taxonomic resolution and data type (abundance and presence) on results. METHODS: Five stream classification systems were used, two based on hydroregions, one based on ecoregions by FEOW, a fourth one based on stream orders and the last one based on clusters of environment variables sampled in 37 streams at Rio Grande do Sul state, Brazil. We used a randomization test to evaluate differences of richness and abundance, a db-MANOVA to evaluate the differences of species assemblages and Classification Strength (CS) to evaluate the classifications performance. RESULTS: There were differences of richness and abundance among stream classes within each stream classification. The same result was found for community data, except for stream order classifications in family level. We observed that stream classes obtained for each stream classification differed in terms of environment variables (db-MANOVA). The classification based on environment variables showed higher CS values than other classification systems. The taxonomic resolution was important to the observed results. Data on genera level presented CS values 12% higher than family level for cluster classification, and the data type was dependent on the classification system and taxonomic resolution employed. CONCLUSION: Our results indicate that classifications based on cluster of environment variables was better than other stream classification systems, and similar results using genera level can be obtained for management programs using family resolution in a geographical context similar to this study.

Inducing Interpretable Voting Classifiers without Trading Accuracy for Simplicity: Theoretical Results, Approximation Algorithms

Recent advances in the study of voting classification algorithms have brought empirical and theoretical results clearly showing the discrimination power of ensemble classifiers. It has been previously argued that the search of this classification power in the design of the algorithms has marginalized the need to obtain interpretable classifiers. Therefore, the question of whether one might have to dispense with interpretability in order to keep classification strength is being raised in a growing number of machine learning or data mining papers. The purpose of this paper is to study both theoretically and empirically the problem. First, we provide numerous results giving insight into the hardness of the simplicity-accuracy tradeoff for voting classifiers. Then we provide an efficient ``top-down and prune'' induction heuristic, WIDC, mainly derived from recent results on the weak learning and boosting frameworks. It is to our knowledge the first attempt to build a voting classifier as a base formula using the weak learning framework (the one which was previously highly successful for decision tree induction), and not the strong learning framework (as usual for such classifiers with boosting-like approaches). While it uses a well-known induction scheme previously successful in other classes of concept representations, thus making it easy to implement and compare, WIDC also relies on recent or new results we give about particular cases of boosting known as partition boosting and ranking loss boosting. Experimental results on thirty-one domains, most of which readily available, tend to display the ability of WIDC to produce small, accurate, and interpretable decision committees.