The peculiarities of the task of authors identifying and determining author's contribution to publications in digital bibliographic codes are considered. The features of the problem of insufficient identification are manifested in the repetition of information, doubling, the presence of authors with completely coincidental names, self-quotation, autoplagiate and plagiarism itself. It is proposed to use publication information that has already been accumulated in the digital library in the form of related object area data and a variety of target thesaurus data, as the author and user of the library. This information contains links whereby keyword contexts, multiple co-authors, and term associations in dictionaries and thesauruses can be used to identify authorship. It is important that an array of scientific publications is considered, since they have an established traditional structure, which allows comparing fixed text elements (annotations, keywords, classifier codes, etc.). Thus, even if the names in the publications are fully matched, the question of authorship can be raised if the publications in the digital library correspond to different subject areas. Resolution of such contradictions is accomplished by evaluating a plurality of links of all elements of secondary publication information. The result of the comparison could be the addition of the author to a specific area, i.e. the extension of the addressee's thesaurus and the author's personal thesaurus, or the appearance of full namesakes in the library, but from different areas of knowledge. It has been shown that modern data analysis tools allow you to evaluate the author's contribution to publication, despite the fact that of course, only the scientific community can evaluate the real contribution to scientific research.
The problems of determining who deserves to be the author of a scientific article and
what is his contribution to the collective publication, if there is no reliable information
in the digital collection, are resolved in various ways. Basically, a comparison of
related articles and a survey of registered authors is performed, as in ResearchGate.
Almost all digital resources known today face the issues related to the identification
of authors in bibliographic systems. When the information is updated, a
"controversial" author, a full namesake, an "old" author with a different transcription in the
spelling of the surname, etc. may appear. Everyone knows the difficulties of their own
identification even in such authoritative databases as WoS and Scopus, when, despite
all the filters set, we get as a result of a search a list of a "mixture" of their own and
other people's works, which is reflected, for example, in publication [
There are a number of requirements for articles and authors in certain specific
subject areas, and they were approved, for example, for authorship in medical research,
but later became generally accepted. An author is someone who participates in the
development of an idea, collection and analysis of data, writing a work, and making
relevant and ideologically justified changes to the text. Nevertheless, these tools are
not enough to determine the author's contribution to collective research, as indicated,
for example, in [
The level of reliability, transparency and documentation of data about authors has
changed. Thus, the problem of authorship is posed wider, and is not limited to
secondary information when indexing in databases. This problem includes the human
factor, that is, interviewing experts, editors and co-authors. In general, there has been
a tendency towards an increase in the number of coauthors over the past 30 years [
This paper discusses the options for using the data that are available in the arsenal of modern information technologies of semantic libraries for indexing publications, authors and their contribution to collective work.
The structure of a scientific publication is a feature of scientific articles that is quite
well-established for many domestic and international journals. The strictness that
authors are encouraged to adhere to in accordance with the instructions from the
publishers is dictated to some extent by the process of digitizing publications for their
subsequent indexing in bibliographic databases. In the 70s of the last century, a family
of standards for machine-readable cataloging (MARC) [
Thus, a list of required fields of secondary information about the document "scientific article" was formed: author, affiliation of authors, title, keywords, classifiers (MSC, UDC and / or specialized), output data (publisher, pages, year). In the future, an abstract, a list of cited literature and identifiers such as ORCID, etc. were added. All these fields are used for indexing publications and can be used as search fields when forming a request and identifying authors.
The difficulty arises if this information is not enough, or it is not in full in the database, or if the user has not it. Refinement is carried out through expert knowledge or through semantic links, which can be implemented in the form of hints from the database.
The body of the publication, as a rule, is not searchable, even if the publication is in the public domain, but is available to publishers for preliminary lexical, syntagmatic, paradigmatic, and semantic processing when placed in bibliographic databases. 2.2
The concept of “addressee in the information environment”, formulated for the
convenience of identifying users and authors from databases, implies a person - a
participant in the information process, search and exchange of information. The term
“addressee (individual) thesaurus” (TA) was introduced into computer science by
Yu.A. Schrader [
The composition of the data (information) of the addressee's thesaurus depends on the conceptual reserve of the individual. For a semantic library, you can focus on the following data set: the frequency vocabulary of the individual; variants of combinations of terms; contexts of frequency terms; special designations and formulas; lists of cited literature; lists of citing authors; list of publications with cross-references. If the information system contains enough data and publications on a certain subject area, then on the basis of the set of data about the addressee's thesaurus and metric analysis, it is possible to build a dictionary-thesaurus of the author's subject domain. Further, comparing subject thesauri, it is possible to more accurately identify their authors, as well as establish the belonging of the text to a certain author and his contribution to research. 2.3
Methods of text comparison for attribution are considered, such as frequency
algorithms [
Using this set of methods, it is possible to form an information processing technology for newly received data in an information bibliographic system.
The first stage of preprocessing publications for each author includes: - frequency processing of texts to obtain a list of terms with their weight (frequency of use); - compiling a list of co-authors; - forming the set of contexts for terms.
As a result, the following data (parameters) of the author are accumulated: list (dictionary) of terms, rank (weight) of terms, word forms of terms, relative frequency of terms (in relation to other terms), absolute frequency of terms, concordance dictionary (dictionary with contexts), Fig. 1. At this stage, it is also possible to select a list of unique terms, designations, formulas and other features of the text, typical for some authors and subject areas.
Author's publications List (dictionary) of the author's terms Word forms of terms
Relative frequency of terms Absolute frequency of terms
The second stage consists in the procedure for comparing authors according to the available (accumulated) parameters. The intersection of the sets of terms, contexts and unique terms, designations, etc.
After comparing and identifying a set of publications belonging to a specific author, an author's index and an index of cited publications are compiled. At the same time, it is possible to vary the severity of the belonging of "controversial" publications to one or another author, taking into account the degree of coincidence of the revealed parameters (in%, for example).
At this point, the preprocessing of the newly received data about the author can be completed. The whole set of related information obtained can be attributed to the addressee's thesaurus.
Note 1. If a series of publications of one group of authors is supposed to be loaded into the system, then at the preliminary stage of processing it is possible to compose a thesaurus of co-authors.
Note 2. If a single work has been received, then preprocessing (according to the scheme in Fig. 1) is used to be included in the existing author index, or in the absence of matches and questionable properties of the publication (variants of surnames and other secondary documents), it is stored in the confirmation status, but participates in further subject semantic processing. Confirmation can be done automatically if the system accumulates additional information about the author or upon request to the author.
For further semantic processing of publications, it is necessary to use dictionaries (thesauri) of professional terms from SD (for example, mathematical SD). Publications must be indexed in accordance with the subject and thematic focus, determining the belonging of the terms of publications to dictionaries (thesauri) of subject areas. Thus, to fix the links of the thesaurus of the addressee (author) with SD. These links are further additional features for the subject identification of the author. As a result, publications that are semantically linked in ontologies, after preprocessing, will have a number of author identification features. 3
Using the example of a number of works in the fields of higher mathematics, we can consider options for identifying authors of publications with similar sets of secondary documents.
For text processing, a free library for high-performance full-text search Apache Lucene, implemented in the Java language, is used. 3.1
To highlight the meaningful expressions of the document, the calculation of the tf-idf
measure was used for the terms of the document extracted from the index, taking into
account the morphology [
author
Иванов identification interface candidate evaluation meaningful values full-text index metadata database
Further, terms for which the tf-idf measure was less than the threshold were
excluded. Composing combinations of two and three words was performed based on the
use of the context of the selected words, and rules that take into account morphology.
The context is understood as N words in the text before the word for which the vector
is constructed, and N words after this word. To highlight the context, a shallow neural
network model word2vec [
As an example, further in Fig. 3 shows the stage of formation of the thesauri of subject areas of individual authors (Russian), on the basis of which one can reason about their (authors') identity.
The example (Fig. 3) shows that the works of the authors were obtained with an incomplete set of secondary information. Application of the described algorithm makes it possible to identify terms, relationships and intersections of subsets of terms, taking into account their contexts.
Further, we use additionally the connections of terms from the encyclopedia, classifiers UDC, MSC and other works from the field of analytical spaces, such as shown in Fig. 4.
MSC 46Axx Topological linear spaces and related structures
Frechet space
Topological vector space Locally convex space
Köthe space
Further, we use additionally the connections of terms from the encyclopedia, classifiers UDC, MSC and other works from the field of analytical spaces, such as shown in Fig. 4.
About 5000 authors of publications were processed. Separately, work is being done to process formulas and include them in the author's thesaurus. A formula comparison algorithm based on a vector model is used. The algorithm is conventionally divided into two parts: the initial selection of candidate formulas and their subsequent ordering by similarity. A description of this algorithm is beyond the scope of this article. 3.2
To take into account the author's contribution to the publication, it is required to
investigate the history of the author's work and his affiliation to scientific schools, as
well as the author's research in subject areas. This is of particular importance as
coauthorship has become commercial in nature and paid publications, “senior
authorship” and citation have become possible [
The set of "historical" data about the author and publication is formed on the basis of the addressee's thesaurus as follows. The history of publications is collected and stored: co-authors, cross-references, keywords, internal system indicators of publications belonging to subject areas (LibMeta).
Almost all modern bibliographic collections collect and display the listed data. The development of information technology makes it possible to use various methods of analysis to establish the authorship of publications. It should be noted that for works of fiction, such as "pocket" detectives, such an examination has been carried out for a long time, since this process was originally built on a commercial basis and it is necessary to take into account the contribution of each participant. The scientific community needs to avoid this approach, as it inevitably leads to downplaying the value of research work.
The criteria by which the publications are distributed are selected, these are the history of the issue, novelty, the number of publications on related topics, many coauthors, expert opinion expressed in the process of discussions and peer review.
Publication history. The structure of a scientific article assumes the presence of an introduction, which lists previous research. Analyzing this text, it is possible to compile lists of researchers and corresponding bibliographic references on the selected topic, example Fig. 5.
Next, select the intersections within these sets and identify the "main" authors and their co-authors. For co-authors to identify frequency characteristics and belonging to the SD. Thus, to obtain a "map" of publications on the topic, where there will be areas of intersection of authors' collectives, where {k1, k2,…, kN} of authors intersect (k1> k2>…> kN). Authors from citation lists are also included in these areas. Individual authors (A, B, C, ...) can belong to many “invitees” to participate in publications, and then their role is evaluated by experts from the scientific community. These can be authors of publications without co-authors working in a given subject area, and then, naturally, their contribution to the work is not disputed.
Subject Domain
Publications
Author A Author B
Author C
The set of authors {k1}, which is larger than others, can apply for many leading scientists, leaders of scientific schools and research projects (grants, etc.).
Keyword scoring. The intersection of keywords in the authors' thesauri indicates the closeness of the studies.
Keyword novelty score. The analysis of the collectives that make up the sets {k1, k2, ..., kN} allows us to identify “new” members of the group of authors, for a certain period of time, “new” keywords for the same period of time. Since, thanks to the TA, it is possible to find out to which author the “new” keywords belong, it can be concluded that thanks to whom the “new” contribution to publications and to the SD.
Metric for evaluating author's participation. Based on the data of the TA, the LibMeta system has introduced a metric for assessing the author's participation in publications in mathematical SD.
The following sets of metrics are calculated for the ODE domain thesaurus: - core of key concepts of the SD (Concept Kernel) – {СK=K1∩K2∩K3 }, where K1 set from ODE domain thesaurus, K2 set from special function dictionary and K3 set from mathematical encyclopedia;
|KK|=|K1|+|K2|+|K3|, |K1|=184, |K2|=151, |K3|=6263, |KK|=6 598; - core of information object keywords for different types of resources of the domain resources (Keyword Kernel) – {KK}, |KK|=6810; - core of authors' collectives by years (Kernel of Copyright Teams) – {KCT}. Consider, for example, 2015, for publications dealing with Bernoulli ODE1 We get: |KK2015|=754, KCT2015={'Лазарев', 'Неустроева', 'Шишкина', 'Бочкарев', 'Лекомцев', 'Сенин', 'Янковский', ‘Кольцун'};
- core bibliographic references (Bibliographic Reference Kernel) – {BRK} for these authors are represented by 34 references, |BRK|=34.
Next, the intersection of data from the author's TA is estimated: - keywords {KWA}, |KWAЛазарев|=14, |KWAЯнковский|=79; - co-authors {CA} |CA2015|=163; - bibliographic lists {RL} |RLЛазарев|=3, |RLЯнковский|=16, with the sets {KK2015}, {KCT2015}, {BRK2015} to the general characteristics of the SD:
{KWA}∩{KK}, {CA}∩ KCT}, {RL}∩{BRK}.
Based on these sets, estimates of the author's contribution to the SD are introduced KWAЛазарев/KK2015=14/754, “average” author's contribution to the SD this year CA2015/KСT2015=163/8, “average” author's contribution |RLЛазарев|/|BRK|=3/34, |RLЛазарев|/|BRK|=16/34.
These estimates show the author's contribution to the SD and to specific research (publications) "over time". We emphasize that these estimates do not reflect the picture of the real world, but they are valid for characterizing the set of objects that are loaded into the system.
In reality, it can be difficult to draw a line between authorship claims, and
sometimes it is a matter of controversy among academic schools. There are cases when an
idea and its implementation in research belongs to different people who may or may
not know about each other's works. This raises issues of plagiarism and priorities in
science. An example of this is the history of disagreements between Newton and
Lebniz on the contribution of each to the development of mathematical analysis [
Authors who have the highest percentage of "overlaps" with the SD of ontology can be considered “key” researchers in the SD.
Note 4: Our study does not provide any assessment of the rationale for the authors' studies and the quality of scientific papers.
Note 5: All assessments are made only on the basis of publications, secondary information or full texts (if available) and author's methods of tracking links in the semantic library.
Note 6: The real contribution of the author to publication and research can only be assessed by the scientific community. In a digital library, you can only set the number of links according to the selected characteristics and on the basis of the data array that is already in the library. This gives a picture of the contribution of the publication and the rating of the author in the scale of the available data, rather than the quality of the publication and the knowledge of the author in general.
Note 7: The LibMeta library has a technology for creating a subject author's thesaurus and on its basis you can get an idea of the addressee's thesaurus as a participant in the exchange of information in the information environment. This technology allows us to consider the meaning and contribution of the author's publications in relation to various subject areas that make up the intersection of sets within the author's subject thesaurus. 4
The technology of preliminary processing of publications for further placement in the digital library is proposed. Using the data of the addressee's thesaurus allows accumulating structured information about authors and publications, which helps to identify authors at the preliminary stage and evaluate their contribution to research.
An ideal scheme for assessing the role of the author and attribution is presented, and of course, there are controversial factors in it, but it can be used as a first approximation if the authorship of the article is in doubt due to the inaccuracy of secondary data in the digital library. However, it is in digital libraries that you can take into account, if not all, but many of the attributes of authorship, which is shown in the examples of mathematical articles in LibMeta.
The use of a personal environment for scientific research on the basis of individual bibliographic collections and the results collected by the author in the process of research allows us to consider the problems of identification and determination of the author's contribution as part of the functioning of the semantic library.
This work was carried out with partial support from the Russian Foundation for Basic Research, project #20-07-00324 and within the framework of the topic ‘Mathematical methods for data analysis and forecasting’.