In the given article content lifecycle support method in Web systems is proposed. The uniform method realizes information resources processing in the Web systems and simplifies the information technology of Web content automation management. In the given paper the main problems Web content function and management services in Web systems are analyzed. In this article, the authors proposed to apply the technique of content analysis of textual information in e-commerce systems for the automation of e-business and decision making by the appropriate person.
The current level of complexity of tasks solved through e-commerce systems is
constantly growing: from the automated information collection and interpretation to the
management, design, modeling and forecasting of the various business processes, for
example, forecasting by the e-commerce system the changes in demand for a certain
type of products depending on the thematic discussions by forum users or the analysis
of arrays of comments on these products. However, the most complex tasks are not
solved automatically by e-commerce systems, but by persons who accompany the
process of the corresponding system [
The article purpose is to design a typical architecture for content management
systems. The work feasibility lies in the methods development of Web resources
processing in such systems. It promotes the active development in Ukraine e-business
and effective implementation of the Web systems [
business globalization increases the demand for Web content and fast access to it; uneven functioning business processes according to the regions. leads to increasing needs on speed, regularity and frequency of obtaining of the necessary Web content; saving time in the necessary Web content getting; personalization in services provide in the Web systems; the Web systems integration. Advantages and feasibility of the
receiving; the cycle reducing of Web content production and sale; costs reducing
associated with the content exchange; the openness of the Web systems due to
customers; automatically inform users of Web content; alternative sales channels creation,
such as online newspaper. Problems of Web resources processing in the such systems
associated with increasing content amounts on the Internet; the rapid prevalence of
Internet access; active e-business development; the set expanding of Web services and
goods; the growth in demand for Web services and products; Internet technology and
facilities creating and applications expansion of Web resources processing techniques
[
Content streams number is greater than the ways number of goods moving to industrial enterprises. Much of the content flows consists easily formalized and automated procedures. The main problem is the common approach lack to process modeling, design and development of Web systems. There is a general and detailed classification lack of Web systems. This leads to the problem of the general methods definition for the architecture and algorithms development of these systems functioning. The existing Web systems do not support the entire content life cycle, do not solve information resources processing and content management problems (Table 2). This justifies the research purpose, relevance, appropriateness and directions.
pointment and conditions. They define the main purpose of modeling. They also make it possible to formulate the requirements for the system formal model S and content management models. Model of Web systems presented as
tem, C c1, c2,, cnC is influences the content flow on the system, V v1, v2,, vnV is the environment influence,
parameters, Y y1, y2,, ynY is the system output characteristics, T t1, t2,,tnT is the content management transaction time. The process S of Web systems functioning described by the function as y j ti t Function xi , cr , vl , hk , ti , where xi is the visitor/user query to the system. Characteristics component y j according to Google Analytics are the visits number for the time period t , the average Time Spent Online (min:s), bounce rate (%), achieved the goal, dynamics (%), total viewing pages, page views number per visit, new visits (%), total unique visitors, traffic sources in % (search engines, direct traffic or other sites). 5
During the investigation of the mechanisms of textual information generation, it was
revealed that the choice of the text information description model depends on how the
probabilistic and linguistic test is constructed and how the extraction of the certain
units is made from the text information [
Content analysis is one of the known analysis methods for the textual
information. It is the standard method of research in the social sciences, the subject of which
is the analysis of the textual arrays content and communicative correspondence
products (for example, comments, forums, electronic correspondence, articles, etc.).
Content analysis is the quantitative and qualitative analysis of textual information and
textual arrays for the purpose of the subsequent meaningful interpretation of the
obtained quantitative and qualitative regularities. Content analysis is used to study the
sources that are invariant in structure or content, but which exist as a
nonsystematized, randomly organized textual material [
topics, and messages that focuses on a message content. So, before analyzing the selected linguistic units, it is necessary to predict their content and determine each possible observation result in accordance with the researcher expectations. As first step in conducting a content analysis of this type, the researcher has to create a dictionary in which each observation will be defined and assigned to appropriate category.
Qualitative content analysis (structural) is a research method in which the
researcher is not so much interested in the message content as in the form and structure
of this message. For example, you need to determine how much time or print space is
allocated to a topic in a particular source, or how many words or columns have been
given to each topic of the corresponding category. Then the relative density of each
topic and category are calculated, and a comparative analysis of the corresponding
topics is carried out for the further prediction of events, processes, and the like. The
relative density P of any category is calculated by the formula: P=R/T, where R is the
number of the analysis units that sets this category, T is the total number of the
analysis units. In [
formalize the division of the whole texts or the selection of their certain items for the subsequent collection.
cal statistics.
According to the Ukrainian researcher in [
structions for the encoder;
The construction of the information analysis systems largely depends on the availability and compliance with the rules for the above steps: 6. The procedure of content-analytical selection should consist of precisely defined actions, which, without any changes, should be subject to all processing objects;
obtained; conclusions should not be based on any part of the results, but rather take into account all of them without exception.
For the automated processing of textual information is of great importance not only what the occurrence frequency of a particular category of the linguistic unit, but generally its presence in the test text or absence. Quantitative estimation makes it possible to draw the objective conclusions about the material focus concerning the number of uses of the analysis units (key quotes) in the studied texts, such as the number of positive/negative responses to a certain type of product. Qualitative analysis allows us to draw the objective conclusions, where there is a certain linguistic unit in the text and in what context. 6
The main stages of content analysis in e-commerce The main stages of content analysis of textual information in e-commerce systems are the following items.
1. Identification of an array of sources under investigation or messages through a set of specified criteria to which each message corresponds: a given source type (forum, email, online newspaper, chat, online magazine); one message type (article, e-mail, banner, comment); specified parties involved in the communication process (addresser, addressee, recipient); confronted size of messages (minimum scope or length); message occurrence frequency; message distribution method; message distribution place; time of message emergence and so on.
2. The formation of the sample multitude of messages. In some cases, the entire set of sources identified at the first stage is studied, since the message to be analyzed is often quantitatively limited and available. However, sometimes the content analysis should be based on a limited sample, extracted from a larger array of information.
3. Identification of the linguistic units under analysis, for example, words or topics. The correct choice of the linguistic units to be analyzed is an important part of the entire work. A word is a simple message element. A topic is the other unit, it is a separate statement about any object. There are clear requirements for the choice of the possible linguistic unit to be analyzed: large enough to interpret the meaning; small enough to not interpret many meanings; easily identified; the number of units is large enough for sampling.
If accepted a topic for the unit analysis, the following rules are taken into account: the size of the topic does not go beyond a paragraph; a new topic arises if there is a substitution of goal, category or someone who perceives or acts.
4. Allocation of calculation units, which can coincide with the meaningful units or be specific in nature. In the first case, the analysis procedure is reduced to the frequency occurrence count of the allocated meaningful unit; in the other, the researcher, on the basis of the analyzed material and the research objectives, sets forth the calculation units that can be: the physical length of the texts; the text area filled with the meaningful units; number of lines (paragraphs, characters, text columns); file size and type; a number of pictures with a certain content, plot, etc.
In some cases, researchers use other calculation elements. At this stage of the content analysis, the fundamental value has a rigorous definition of its operators.
5. The calculation procedure itself. In general, it is similar to the standard methods of classification by the selected groups from the formulas of mathematical statistics and the probability theory. There are also special counting procedures for the content analysis, for example, the formula for calculating the Janis coefficient c , designed to calculate the ratio of positive and negative (towards the selected position) evaluations, opinions, arguments. The Janis coefficient can be used, for example, to calculate the ratio of positive and negative thoughts, highlighted in user comments about products that are sold through the e-commerce system. If the number of positive evaluations exceeds the negative ones, the Janis coefficient is calculated by the formula: c f 2 f n , where f is the number of positive evaluations; n is the number of r t negative evaluations; r is the volume of the text content that has a direct relationship to the problem that is being investigated; t is the total volume of the analyzed text. If the number of positive evaluations is less than the negative, the Janis coefficient is found by a formula: c f n n2 .
r t 6. Interpretation of the obtained results according to the goals and objectives of the case study. Usually, at this stage, the characteristics of the text material are identified and evaluated, which allows one to draw conclusions about what the author wanted to emphasize or hide. Or, based on a statistical set of calculated Janis coefficients for a certain period of time for a specific category of products, you can predict changes in demand for the same products. 7
Advantages of content analysis in e-commerce The use of content analysis of textual information in e-commerce systems, according to the authors, provides a number of advantages for simplifying the business dealing and solves many problems facing business process participants. Here are a few basic 1200 1000 800 600 400 200 0 bookclub.ua presstime.com.ua 700 830 720 bonprix.ua timesonline.co.uk 970 advantages of content analysis of the text information in e-commerce systems: the automation of the textual information filtering, which is placed on the website of the e-commerce system by a user; the possibility to automatically create a portrait of a permanent user based on his/her comments; the possibility to automatically create a portrait of the target audience based on the analysis of portraits of permanent users; the reduction of the number of moderators who serve the e-commerce system; the reduction of the time for filling the site with the text information of a permanent user due to the automatic processing of this information and the absence of an intermediate link as a moderator; the language barrier elimination due to the automatic generation of dictionaries of the permanent user and the use of automatic translation.
The average number of system queries
tion as Source content formation content management content implementation database.
Model of Web systems S X , Formation, C, Management, Realization,Y , where X x1, x2 ,, xnX is input data set, Formation is content formation operator, C c1, c2 ,, cnC is content set, Management is content management operator, Realization is content implementation operator, Y y1, y2 ,, ynY is output data set. Below is the content management models classification.
stages connection as Content content editing Database content presentation informational resource. Pages generate model on demand as ManagementQ X , C, Q, R, Edit,Y , where X is input data set, C is content set, Y is pages generated set, Q is query set, R is pages formulation and submission function, Edit – content editing and updating function.
communication as Content content editing database informational resource.
When making changes to the site content creates a static pages set. Not is taken into account interactivity site between visitors and content. Pages generation system model while editing as ManagementE C, Edit,Y , where C is content set, Y is static pages set, Edit is content editing function. The pages formation described func tion as y t Edit c,Weight,t .
is presented as a communications major stages as Content content editing Database content analysis blocks collection content presentation informational resource. This model is as ManagementM X , C,Q, R, Edit, Caching ,Y , where X is input data set, C is content set, Y is pages generated set, Q is query set, R is pages formulation and submission function, Edit is content editing and updating function, Caching is cache formulation function. Cache is update automatically (after a certain period or when amending certain site sections) or manually (team administrator). Another approach is to maintain information blocks on the editing site stage. Then the pages collected from these units when requesting user. The process is implementing caching. The module generates submission page once. Then it is downloading from the cache several times faster.
User queries content analysis allows to qualitatively assess the content flow in the system. This facilitates the subsequent decisions by the moderator as follows: the problem situation description and study purpose search; precise definition of the study object and subject; the object preliminary analysis; concepts substantial clarification and empirical interpretation; the procedures description for the properties and phenomena registration; the overall study plan determining; the definition of the sample type, sources collection and so on. Qualitative content analysis is intended to provide the necessary means moderator for results analysis (Table 3). With their help identify the content properties and test them on the general content stream. Then apply the content stream general properties on its specific thematic part.
Stage name
quantitative evaluations. Next qualify them as categories.
tional and retrospective databases forming; databases rotation; users work personalization; a database create and access to it; personal needs and sources protection; work statistics keeping; search ensuring in database; output forms generation; interaction with databases of other subsystems. In Table 5 presents the content management key stages in the Web systems.
creates a database search primary content pattern (PCP). They used clustering technology (automatic forming groups with similar content on the criteria PCP). In content management systems formed database annotations for used in the search process.
description. Database record is performing automatic abstracting methods (digest is formation of text statistical analysis methods). These methods are used to create the
Content formation Results search and clustering
Content categorization New content stream
Author Traditional database
Cluster: content title, description, number clusters database Centroid choosing,
describe annotations forming content association in group of the likeness
Content database Topics tracking Themed string
Fig. 2. Scheme of the Web systems functioning from annotated database
Thematic content Indexing
Moderator Database creating for retrieval pattern Pattern database annotations database Clustering New themes identify
Load distribution
Content Moderator result clusters set
K K K K ...
K
User
Theme query Quest formation Search procedure Result clustering Result abstracting Reports generating
Report submission SK SK SK ... SK K
K ...
K C ... C
...
Page submission
Resource formation The results display, including the name and description of the super cluster Clusters Content grouping Thematic clusters
C
Clusters grouping of requests considering The cluster name and description print
Output
K content C
C
Personalization based rules is the content provision to specific users or users groups of conventional business logic using. For example, using a rule where all those interested in children's books fall into the group that focused advertising children's clothing. Rules developed on the content basis that type users in a registration card. In content management systems use algorithms categorization with personalization using filters (intelligent agents). Also, algorithms use based on the content analysis of user behavior. In particular, he analyzes the content to which the user accesses, the sites visited and more. That is constantly conducted analyzed registered user and user’s group history with priorities an overwhelming number for interest.
The full-text search problem in large content arrays is ineffective. The annotated content search solves the problem exactly: instead of the full content searching to search on annotations (pattern). Digest remotely similar content and often not perceived by person. But as the search content pattern with weighted keywords and phrases it leads to adequate results with full-text search. Digest constructed from content fragments with the largest weight values. Content analysis is used for digests automatic generation, the concepts (categories) relationship automatic detection, relationships automatic clustering to the most important identify, the relationships automatic detection (e.g., positive and negative). One of the most important tasks in content analysis is the categorization process. It sets the conceptual grid. In its terms is the content flow analyzing and new categories generate. In Fig. 2-6 presented the results of the developed Web systems “Good morning, accountant” (dobryjranok.com).
1 5 9 13 17 21 25 29 1 5
Ukraine (not set) Russia
Austria
Italy
Norway Poland Fig. 4. Resource visit distribution from different countries for the 07-12/2012 period 200 150 100 50 0 ru ru-ru uk en-us de en en-gb uk-ua доброго ранку бухгалтер dobryjranok.com
іт пільги з пдв добрий ранок, бухгалтер Fig. 7. Resource visit distribution of the search engines under the keyword content-search for the 09-12/2012 period
cessful marketing studies is use statistical basis. Globalization is more than a simple Web-pages conversion from one language to another. It still is to promote local brands (localization). Within globalization and localization is using content strategy, information hierarchy and navigation structure. For appropriate systems implement the following tasks: create content in languages that its list is predefined; global/local content preservation in different databases; support services for text automatic translation; content changes track and its approved.
In the article the justification is needed in the information resources processing methods and means development in electronic commerce content. This is implemented by the architecture improving of Web systems. Improving such systems is to automate processes of the Web content formation, management and implementation. This paper is analyzes the terminology and classification of Web systems. This is done to determine the characteristic design and simulation patterns, trends, process of Web systems. Also, the article outlines the shortcomings of existing content management methods and means. For Web systems is developed and described a content management model. This made it possible to develop a generic architecture for typical Web systems and standardized methods of information resources processing in these systems. The paper improved the overall architecture of Web systems. It is different from the existing modules presence of information resources processing. This is enabled the stages implementation of the content lifecycle. The content analysis of textual information in e-commerce systems allows us to determine the incidence of any feature of the studied set of texts. In such a case, it is important not so much absolute as the relative importance of the feature, that is, the characteristic of the place (share) among other features. For example, the percentage point of discussions by the forum users on the economic issues in relation to the political issues, or the percentage of positive comments on products in relation to the negative ones and with respect to all comments on this product category in the online store. Measuring the ratio between the features in the text gives the empirical material for understanding the functional relationships between the elements of reality reflected in the texts, for example, determining the mood of the audience of Forum on the economic or political situation in the country and the world. If there are chronologically ordered texts, you can have a number of portraits of the studied reality fixed over time (the change in demand for the product category depending on the season, for example, fantastic fiction is read more in the winter, and detective novels − in the summer) or portraits of the target audience (the change in demand for the product category according to the gender, for example, the demand for women's clothing is greater in September than in March). It allows you to hypothesize about the predictive nature of system elements functioning.