Professional Value of Scientific Papers and Their Citation Responding

Abstract

In the course of the last thirty years, science enjoys a remarkable quantitative boom. For example, the total number of substances, registered in the Chemical Abstracts Service Registry File (CAS RF) at the end of the year 1985, was about 8 millions while at the end of the year 2015 it reached up to 104 millions. But, still more and more behind this quantitative boom of science are some of its qualitative aspects. So, e.g., the x–y–z coordinates of atoms in molecules are presently known for no more than 1 million of substances. For the majority of substances registered in CAS RF, we do not know much on their properties, how they react with other substances and to what purpose they could serve. Gmelin Institute for Inorganic Chemistry and Beilstein Institute for Organic Chemistry, which systematically gathered and extensively published such information since the nineteenth century, were canceled in 1997 (Gmelin) and 1998 (Beilstein). The number of scientific papers annually published increases, but the value of information they bring falls. The growth of sophisticated ‘push-and-button’ apparatuses allows easier preparation of publications while facilitating ready-to-publish data. Articles can thus be compiled by mere combination of different measurements usually without idea what it all is about and to what end this may serve. Driving force for the production of ever growing number of scientific papers is the need of authors to be distinguished in order to be well considered in seeing financial support. The money and fame are distributed to scientists according to their publication and citation scores. While the number of publications is clearly a quantitative criterion, much hopes have been placed on the citation, which promised to serve well as an adequate measure of the genuine scientific value, i.e., of quality of the scientific work. That, and why these hopes were not accomplished, is discussed in detail in our contribution. Special case of Journal of Thermal Analysis and Calorimetry is discussed in more particulars.

Quo usque tandem, scientometrics? or voice in the desert

Appendix: Factor Analysis

Spectra (pattern vectors of identification features) $\vec{z}_{1} ,\vec{z}_{2} , \ldots ,\vec{z}_{k}$ of components and their abundances in an analyzed mixture can be reconstructed (synthesized, extracted) from the spectra of this mixture $\vec{x}( \equiv \vec{x}_{1} )$ and its (p − 1) fractions $\vec{x}_{2} ,\vec{x}_{3} , \ldots ,\vec{x}_{p} (p \ge k),$ which can be obtained by separation of the mixture under consideration [60, 61]. Expressing the vectors $\vec{x}_{1} ,\vec{x}_{2} , \ldots ,\vec{x}_{p}$ by linear superposition of transposed eigenvectors $\vec{q}_{1}^{\prime } ,q_{2}^{\prime } , \ldots ,\vec{q}_{n}^{\prime }$ of the Gramian matrix ${\mathop{X}\limits^{\frown}}^{\prime}\mathop{X}\limits^{\frown}$ of the data matrix

$$ \widehat{X} = \left[ {\begin{array}{*{20}l} {\vec{x}_{1} } \hfill \\ {\vec{x}_{2} } \hfill \\ \cdots \hfill \\ {\vec{x}_{p} } \hfill \\ \end{array} } \right] = \left[ {\begin{array}{*{20}l} {x_{11} } \hfill & {x_{12} } \hfill & \ldots \hfill & {x_{1n} } \hfill \\ {x_{21} } \hfill & {x_{22} } \hfill & \ldots \hfill & {x_{2n} } \hfill \\ \ldots \hfill & \ldots \hfill & \ldots \hfill & \ldots \hfill \\ {x_{p1} } \hfill & {x_{p2} } \hfill & \ldots \hfill & {x_{pn} } \hfill \\ \end{array} } \right] $$

associated with the eigenvalues $\lambda_{1} ,\lambda_{2} , \ldots ,\lambda_{n}$

$$ \widehat{X}^{\prime } \widehat{X}\vec{q}_{j} = \lambda_{j} \vec{q}_{j} ,\quad j = 1,2, \ldots ,n, $$

arranged in descending order $(\lambda_{1} \ge \lambda_{2} \ge \cdots \ge \lambda_{n} )$ then

$$ \vec{x}_{i} \sum\limits_{j = 1}^{n} {u_{ij} \sqrt {\lambda_{j} } } \vec{q}_{j}^{\prime } ,\quad i = 1,2, \ldots ,p. $$

From this it is obvious that eigenvectors associated with the largest eigenvalues are most important and eigenvectors associated with the smallest eigenvalues are least important. So, retaining only the first k eigenvalues, which are significant at an accepted level, we have for $i = 1,2, \ldots ,p$

$$ \vec{x}_{i} \doteq \sum\limits_{j = 1}^{k} {v_{ij} \vec{q}_{j}^{\prime } }. \quad {\text{Due to the fact that}} \;\;\vec{x}_{i} = \sum\limits_{j = 1}^{k} {d_{ij} \vec{z}_{j} } $$

it holds, within the same tolerance, as well

$$ \vec{z}_{j} = \left[ {z_{j1} ,z_{j2} , \ldots ,z_{jn} } \right] \doteq \sum\limits_{m = 0}^{k} {t_{jm} \vec{q}_{m}^{\prime } ;\quad j = 1,2, \ldots ,k.} $$

The coefficients $t_{jm}$ (elements of the matrix $\mathop{T}\limits^{\frown} = \left[ {t_{jm} } \right]_{j = 1, \ldots ,k}^{m = 1, \ldots ,k}$) are then determined making use of the fact that intensities of lines of the components of analyzed mixture in the spectra of the mixture and its fractions are nonnegative

$$ \widehat{Z} = \left[ {\begin{array}{*{20}l} {\vec{z}_{1} } \hfill \\ {\vec{z}_{2} } \hfill \\ \ldots \hfill \\ {\vec{z}_{k} } \hfill \\ \end{array} } \right] = \left[ {\begin{array}{*{20}l} {z_{11} } \hfill & {z_{12} } \hfill & \ldots \hfill & {z_{1n} } \hfill \\ {z_{21} } \hfill & {z_{22} } \hfill & \ldots \hfill & {z_{2n} } \hfill \\ \ldots \hfill & \ldots \hfill & \ldots \hfill & \ldots \hfill \\ {z_{k1} } \hfill & {z_{k2} } \hfill & \ldots \hfill & {z_{kn} } \hfill \\ \end{array} } \right] = \left[ {\begin{array}{*{20}l} {t_{11} } \hfill & {t_{12} } \hfill & \ldots \hfill & {t_{1k} } \hfill \\ {t_{21} } \hfill & {t_{22} } \hfill & \ldots \hfill & {t_{2k} } \hfill \\ \ldots \hfill & \ldots \hfill & \ldots \hfill & \ldots \hfill \\ {t_{k1} } \hfill & {t_{k2} } \hfill & \ldots \hfill & {t_{kk} } \hfill \\ \end{array} } \right] \cdot \left[ {\begin{array}{*{20}l} {\vec{q}_{1}^{\prime } } \hfill \\ {\vec{q}_{2}^{\prime } } \hfill \\ \ldots \hfill \\ {\vec{q}_{k}^{\prime } } \hfill \\ \end{array} } \right] = \widehat{T} \cdot \widehat{Q} \ge \widehat{O} $$

where

$$ \widehat{Q} = \left[ {\begin{array}{*{20}l} {\bar{q}_{1}^{\prime } } \hfill \\ {\bar{q}_{2}^{\prime } } \hfill \\ \cdots \hfill \\ {\bar{q}_{k}^{\prime } } \hfill \\ \end{array} } \right] = \left[ {\begin{array}{*{20}l} {q_{11}^{\prime } } \hfill & {q_{12}^{\prime } } \hfill & \cdots \hfill & {q_{1n}^{\prime } } \hfill \\ {q_{21}^{\prime } } \hfill & {q_{22}^{\prime } } \hfill & \cdots \hfill & {q_{2n}^{\prime } } \hfill \\ \cdots \hfill & \cdots \hfill & \cdots \hfill & \cdots \hfill \\ {q_{k1}^{\prime } } \hfill & {q_{k2}^{\prime } } \hfill & \cdots \hfill & {q_{kn}^{\prime } } \hfill \\ \end{array} } \right], $$

and also abundances $d_{i,j}$ of those components in the analyzed mixture and their fractions are nonnegative

$$ \widehat{D} = \left[ {\begin{array}{*{20}l} {d_{11} } \hfill & {d_{12} } \hfill & \cdots \hfill & {d_{1k} } \hfill \\ {d_{21} } \hfill & {d_{22} } \hfill & \cdots \hfill & {d_{2k} } \hfill \\ \cdots \hfill & \cdots \hfill & \cdots \hfill & \cdots \hfill \\ {d_{p1} } \hfill & {d_{p2} } \hfill & \cdots \hfill & {d_{pk} } \hfill \\ \end{array} } \right] = \widehat{X}\widehat{Q}^{\prime } \widehat{T}^{\prime } \left( {\widehat{T}\widehat{Q}\widehat{Q}^{\prime } \widehat{T}^{\prime } } \right)^{ - 1} \ge \widehat{O} $$

(the Procrustes problem of quadratic programming).