Keywords

1 Introduction and Motivation

The ubiquity of the World Wide Web and the development of large-scale distributed web services promote a dramatic growth in the amount of accessible data through the web and introduces new challenges for end users. Due to the large amount of online data sources, end users are now – in principle – be able to access and integrate the relevant exact information. On the other hand, web services generate data in an ad hoc manner; hence, the systematic management of data has become an obstacle for efficient knowledge extraction and decision making.

For non-expert users, who do not have programming skills, information management and data integration is very complicated, and hinders them to make use of the full potential of the provided data and services for decision making. End User Development (EUD) is a research field with focus on techniques and methods to support the development needs of non-expert users. In the domain of Web Engineering EUD concentrates on empowering non-expert web users to create web applications, which satisfy their need [1]. Expertness in here refers to web technology knowledge in other words non-expert users can be domain or technical experts without web engineering knowledge.

The problem that we aim to tackle in this PhD research is the lack of suitable tool support to enable end users in systematically achieving their data-driven decision-making goals. Integrating and analyzing the huge amount of data from different sources to create new functionalities and values demand sophisticated skills, which non-expert users lack [2].

One of the promising approaches in EUD domain is Mashup technology [3]. Mashup tools have been introduced as a solution for achieving new functionalities by combining various data resources and services across the web [4]. The emergence of mashup technics empower non-expert end users to develop their own application [5]. However, even in semi-automatic mashup tools, users still need background knowledge regarding web technologies and programming. In the semi-automatic mashup approach, end users manually select data sources and services. This selection does not only require knowledge about data source locations, but also about the semantics and structures of data. Moreover, to fulfill the reliability and security requirements end users should be aware of all kinds of related data source aspects; otherwise, due to lack of sufficient information, decision quality might be affected negatively [6]. Another limitation of this approach is a lack of adequate flexibility for more dynamic scenarios where the data providers change their structure or behavior. Therefore, end users need a more flexible and systematic platform, which enables them to create their own tools using the big data produced by the Internet of Services (IoS) and Internet of Things (IoT).

A potential platform to remedy the situation should be able to integrate data from various sources and to provide users with a unified view of data. The platform should relieve users from navigating one source to another to access the data. The data is retrieved from different sources with different structures, therefore preprocessing is required. The next step is knowledge extraction from existing sources and visualizing the result.

A potential platform should fulfill the following requirements. These requirements are divided into two groups, namely framework requirement and End user requirement.

  • Framework requirements:

  • Extracting, integrating and analyzing relevant data

  • Flexible enough to handle data from various domains of knowledge to provide domain specific solutions

  • Flexible enough to handle data from different sources such as social streams, open APIs and IoT streams

  • Ability to work in heterogeneous, distributed and dynamic environment

  • Ability to represent information in visual component

The following requirements are extracted from [7] as some general user interface design principles:

  • End user requirements:

  • Easy to learn the workflow

  • Using familiar terms

  • Consistent user interface

  • Minimizing the element of surprise for users

The proposed platform is a web-based application, mainly implemented by Python programming language and Flask frameworkFootnote 1. For the visualization part, the d3.jsFootnote 2 library was used. A Deep Learning algorithm provided by genismFootnote 3 python library was used as well to increase the intelligence and efficiency of the solution.

2 Research Issues, Objectives and Questions

In this section an overview of research issues and objectives will be given. The main objective of this project is “to support knowledge workers in decision making with a systematic end user development approach”. Using the envisioned platform enables non-expert users to make informed decisions by spending less effort on data definition, integration and knowledge extraction. Using knowledge extraction and Artificial Intelligence techniques, increases decision precision in the overwhelming data environment.

Some of the issues, which need to be addressed in our solution are compatibility with different domains of knowledge and capability of analyzing the different forms of data to produce knowledge. Other issues we faced while solving the problem are reusability, sustainability and security issues.

This research aims to prove the following statistical hypothesis:

  • H 1 : The average user’s ratings are higher for the envisioned platform compared to other existing mashup platforms.

  • H 2 : the average time required to access the exact information and make a decision is lower compared to the average time needed for the same task without using the platform.

3 Research Methodology and Research Design

In this section, we discuss the used methods and techniques in this research project. One of the used methodologies is Design Science. Based on the definition in [8], Design Science is a research paradigm in which human problems are addressed via innovative artifacts, thereby contributing new knowledge to the body of scientific evidence. In Design Science research the key focus is contribution of innovative and sound knowledge [9].

Another tool to guarantee a clear and organized project management procedure is Logical Framework Approach. This approach is one of the European Commission’s suggestions for project management. LFA is an analytical process which aids us during planning and management phase and involves stakeholder and problem analysis, objective setting and strategy selection. This approach provides a structured way to organize the information so that the weaknesses can be identified and important questions can be asked [10].

Throughout the rest of this section we discuss the necessary steps to accomplish the project. The first step was performing a literature review to identify the existing solutions and requirements. For the next step, problems and objectives were identified and formulated in the form of problem and objective trees. The Logical Framework Matrix was developed to capture the main focus points of research. After identifying the stakeholders and their needs the platform requirements can be derived respectively. To prove the feasibility of the solution and perform the first evaluation, a prototype of the platform will be implemented. The prototype should fulfill the main functional and non-functional requirements. A user journey map is another tool that will be used to gain a better understanding of the interaction between users and the platform. This tool illustrates user’s needs, requirements and expectations. Therefore, it can be used to gain useful information regarding the success of the solution.

To evaluate different platform’s functionalities, we plan to conduct a series of validation phases (levels). For the first level of validation, prototyping will be used. Prototyping will help us to evaluate the platform in the terms of feasibility and functionality. In the next level after developing the first version, a series of user studies will be conducted by different groups of users. At the end of each series the platform will be improved and modified according to the feedbacks and newly found functionalities. The important criteria are usability, flexibility and efficiency. New criteria will be determined throughout the process. The evaluation attempts to answer the following questions:

  • Q1: How usable is the platform from an end user’s perspective with different levels of programming skills?

  • Q2: How much time spent for decision making will the use of the platform reduce and how much effort is required to make a decision?

  • Q3: How precise and intelligent are the decisions?

To evaluate the usability, users should rate the platform on a 1 to 5 Likert scale regarding the complexity and ease of use. Then, the average ratings will be used to assess the usability. In terms of efficiency, the required time and effort to find the information and make the decision will be considered. The effort can be expressed in the number of times the user switches from one data source to other.

4 Preliminary Key Results or Contributions

To address the mentioned problems, we propose a platform for End User Development. The IKEV (Intelligent Knowledge Extraction and Visualization) platform should fulfill the end user’s requirements as well as resolving the drawbacks of previous works. For the clarity’s sake, we use a scenario in which the user can employ the IKEV platform to achieve her goals and later in this section the platform architecture will be introduced.

4.1 Use Case Scenario

A research group is trying to find the best fitting scholarship for funding their projects in the field of computer science. To find the best funding opportunity, researchers should go through all calls for proposals. This procedure requires time and effort to gather information about the funding agencies, compatibility to the group’s field of research and knowledge regarding each member’s interests and skills. The IKEV platform will assist researchers to create their own personalized profile based on their publications and projects. Required information for creating the profile will be extracted from agreed-on data sources such as, ResearchGate, Google Scholar etc. Each profile contains information regarding the member’s research interests and other personalized information. By using this information and data regarding the call for proposals, the platform can choose the best fitting call. This scenario demonstrates the platform’s ability to provide a simple, customized and intelligent solution which saves the user huge amount of time and effort. The involved actors of this scenario are academics who are experienced in terms of searching and using the internet. Due to their experience in academic research they possess higher skills in using web sources compared to normal users. This target group can be of any research discipline other than computer science or web engineering, therefore they might not have skills in using web technologies to create their own customized intelligent solution. Our solution is also applicable for normal users with limited “internet using skills”, i.e. users who are able to conduct normal use of the web, but not programming like tasks. The critical difference between normal and experienced users in this scenario is their ability to identify and access the relevant web sources - but in the terms of web technology knowledge they both are at the same low level. The following Table 1 differentiates the different end users:

Table 1. End users definition
Full size table

4.2 IKEV Platform Architecture

The IKEV platform architecture is presented in Fig. 1. This architecture should incorporate the essential components to fulfill the mentioned requirements and support the end users in our scenario.

Fig. 1.
figure 1

Platform architecture

Full size image

The main sources of information for the platform are web sources, which consist of the social stream such as crowd data, open APIs and IoT streams such the data provided by sensors. The end users use client devices like mobile and web clients to interact with the platform. The main blocks of the IKEV platform are Authentication/Authorization Manager, Planning Assistant Module, Source/Service Editor and Information Store.

  • Authentication/Authorization Manager is responsible for granting access to the users based on their credential and information stored in User’s Profile repository.

  • If access is granted to the user, he can use the User Dashboard to make a new planning, edit existing one or manage the data sources.

  • Source/Service Editor’s main functionality is identifying and providing access to relevant data sources and analyzing the data. The Source Discovery and Data Analysis components are respectively responsible for these functions. Source Compatibility Checker checks whether the data source is reasonable to be used for decision making. This component is important for the normal users with limited experience in using the web sources to help them identify the relevant information. Finally, the Visualization Component provides visual results by using the visualization template stored in Visualization Repository.

  • Planning Assistant helps end users to make a sound decision based on the provided data. The more optimized the plan becomes, the higher the user’s satisfaction that can be achieved. The Logical Component uses the Rules Repository to support end users in making logical decisions. Another component in this block is Tool Kit which enables end users to benefit from several project management tools such as various diagrams, problem and objective tree, user journey map etc. New plans (strategic decisions) can be suggested to the users based on the previous decisions and results from Logical Component.

  • Information Store contains all the required repositories for the platform.

The main expected contributions of the project in the first place, is enhancing usability and efficiency of End User Development platforms for making better decisions in less time with less effort. Additionally, increasing the intelligence and also introducing User Journey-based solution are other contributions of this work.

5 Conclusions and Work Plan

In this PhD project we plan to implement a platform to aid end users with limited knowledge in computer and web programming to achieve their goals. We introduced a scenario for knowledge workers and presented an architecture for a platform to help them.

So far, the literature review and management phase has been finished and the LFA matrix, user journey map, and initial version of architecture were developed. Moreover, the first demo of the platform was also provided. This demo demonstrates the Source/Service Editor block of the architecture which identifies each researcher’s specialty by accessing and analyzing their publications and creates a customized profile for each. This progress can be considered as the first level of knowledge extraction in the project.

The next steps are upgrading the architecture to the final version and developing the remaining components of the architecture. Next step after completing the demo is conducting the evaluations to make sure the requirements are met and the performance is at an acceptable level then the final modifications can be carried out accordingly.

The next prospective paper is associated with the initial version of the demo and the achieved results after the evaluation. Since the blocks of the architecture are distinct and each has a different responsibility, after completing and enhancing each block a new paper can be proposed to show and discuss the result. To illustrate the detailed plan following Gantt chart is provided (Fig. 2):

Fig. 2.
figure 2

Gantt chart

Full size image