This paper describes the national funded project entitled PaloAnalytics, which develops an innovative platform that allows companies and organizations, that operate in several countries, to monitor and analyze, in depth, the markets' interest to their products and successfully plan their marketing and communication strategy, with data and insights collected from all the local media, and focuses on its application to cultural spaces and museums. In this notion, we examine the e ect that this project can have in cultural spaces or companies related to arts and culture. PaloAnalytics platform allows organizations to investigate the impact of their products on consumers across di erent countries and this is achieved with the analysis of content from sites, blogs, social networks and open data. This implies that cultural organization can bene t by adopting the implemented services, so that the can recognize and analyze their audience, their online marketing campaigns as well as examine the impact of their messages and the spread of their messages on the Internet. In this paper, we brie y describe the project and discuss on the impact on cultural related organizations.
The data that is generated daily in the world of the internet is vast. The amount
of information is such that it is impossible for companies and organizations to
fetch, analyze and learn from all the data produced. In this scope, PaloAnalytics
is a project that aims to perform the procedures of collecting, analyzing and
extracting useful information from di erent sources of the internet, web pages,
news portals, open sources, and social media. The procedure of collecting and
analyzing information from diverse sources is not something new, and has
attracted research during the last 20 years [
Market research has proven that companies and organizations are in strong need for an holistic market monitoring and analysis service in several countries and not solely the country of their origin; or at least they are convinced that they can perform much better if they have such a tool. Besides, the competition of such companies and organizations is usually international. Furthermore, it clear that when analyzing data in an international environment each local information can easily a ect the whole organization, but it is usually di cult to become a part of the organization's international policy. In general, it seems possible that data can be collected and analyzed in some extent locally but is usually not transferred as knowledge to the international level. In fact, for such organizations it would be extremely useful to utilize a unique language for all the data analyzed and it seems that the English language is acceptable and consistent. A number of tools have been developed including Mention 3 and Brandwatch 4 in order to collect and analyze data internationally but they have some major disadvantages. They focus mainly on social media and target experienced users, while in parallel they do not provide translations of reports from local languages in a universal language. Furthermore, they do not o er a homogeneous overall picture for all the countries that are of interest for a business.
In this ground, we introduce PaloAnalytics project, which intends to focus on the basic challenges that organizations face and includes the ability to have a universal monitoring tool, with links and interconnections between data collected and analyzed from a number of di erent sources and di erent languages. In this way the project will be the ideal solution for international companies (or companies willing to become international) and companies that their international competition a ect their local business. An ideal solution, through which the organization will be able to get information out of large sets of data.
The proposed design and implementation, introduces a series of software modules that will { analyze multilingual content posted on news sites, social networks and open data { extract knowledge and information about products and companies, including product characteristics { analyze sources, their in uence and trends 3 https://mention.com - Mention: Scour the web, social media, and more for powerful market insights 4 https://www.brandwatch.com/ - Brandwatch: Know what your customers think { help in assessing the image of the business and its products as well as its competitors { visualize the knowledge in order to easily understand the analyzed information
These procedures describe how this project can be used by any type of company. Research has shown that cultural spaces and organizations have started to take seriously the world of the Internet and the Social Media. Consequently, they nd it attractive to spread their messages through these mediums, as it is expected to reach a larger and global audience, they can make serious debates and conversations, and, generally, have an alternative active role in order to challenge the mass culture. Having the aforementioned as a base, it is evident that the project can help all these organizations have a holistic presence in social media and the internet; a presence that can be expected to be international.
The rest of the paper is structured as follows: Section 2 presents the methodology of the project, while section 3 discusses the system architecture. In section ?? a detailed description of each component is presented, providing more emphasis on the Trending Topics software module and its results. Section 4 de nes the expected outcomes of the proposed system and the nal section presents a discussion on the project. 2
Due to the large number of di erent modules, the high complexity of their implementation and the importance in precision of their algorithmic procedures, an advanced methodology is employed. As such, the Rational Uni ed Process is used. It is a software engineering procedure that ensures producing high quality software and achieving end user needs within a speci c timetable and cost. Two cycles of project evolution are followed, one that leads to the basic implementation and is longer, while a shorter one will be done in order to perform re nements. Both of the cycles will go through the same steps of development. During the rst cycle the implementation will be ensured, while the second cycle will focus on the quality of the outcomes. The cycle phases include: { Inception Phase { Elaboration Phase { Construction Phase { Transition Phase
During the inception phase a general description of the key requirements of the project is done; key points and the basic constraints are de ned and the system use cases are de ned in brief. An initial business case including the business framework, the success criteria and nancial forecasting is the ones that lead to the project plan and to a draft business model. While analyzing the information, during the processing phase, the use case model was completed, and the nal requirements were recorded. The architecture reached its nal form and the project's development plan was nalized. Currently the project is under the rst construction phase, where modules are implemented and starting to be integrate into the PaloAnalytics platform. Upon completion of the rst phase of implementations an overall system functionality, performance and usability test will be done.
The development of the platform follows a bottom-up approach, based on the proposed architecture as presented in gure 2), starting from data collection that will directly lead to data aggregation services which will be used individually. On the produced data, multilingual content analysis' services are employed, while in parallel, at this stage, business intelligence extracting solutions are applied. The availability of the proposed services will be both on Web and Mobile application enabling increased penetration into the business community. Each service is built supporting endpoint integration in order to be available for use as an individual component even for third party systems, external to PaloAnalytics platform.
This will develop a complete development stack, that is based on multilingual content from news sites, open data sources and social media. The services of this stack are expected to attract third-party businesses companies, public bodies and researchers who will develop new management modes of business data from the sources incorporated by PaloAnalytics platform and will set up new business models on them, multiplying the bene ts for the companies and organizations
Fig. 2: Proposed architecture while maximizing the in uence of the proposed solutions for the scienti c and business community. 3
According to the architecture presented in gure 2 the proposed system is divided into several components and modules enabling in this way individual design and integration. The system, though, can be separated into four major components: { data entrance point / data storage { deep data analysis { semantics and metadata analysis { point of presentation
Each of the major components consists of a number of modules in order to successfully achieve its scope. Furthermore, each component will o er services for direct data extraction and usage by third party systems. 3.1
The entry point of the system is the component that is responsible for collecting and storing data from the several di erent sources (social media, news and open data). The data storage is built enabling several interfaces to be connected in order to fetch and store data. In general, it follows a hybrid scheme including both an SQL and a noSQL database.
The system acts as a data warehouse, including modules for data extraction, data transformation as well as data loading. The extraction of data is done from several di erent sources including news websites, blogging platforms, social media - focusing on text based ones - and open data sources. The data collected is transformed in order to formulate similar objects with speci c uni ed structure. The uni ed structure of each unique object includes a unique identi er, title, body, source, timestamp and author.
The aforementioned is the main object of the system and described the main form of data collected. A number of metadata and objects analyzing in depth each object is used including detailed information about the source, the author, accompanying multimedia and more. Figure 3 presents a generic schema of the database infrastructure that is used in order to support the essential for the system storage.
The data collected are stored on both an SQL-like storage environment as
well as a noSQL environment. The hybrid scheme will help for storing elements
for fast access in the noSQL nodes and collection of all the collected data in an
SQL based structure for better interconnection between them and permanent
storage of data with historic metadata [
The system core includes the following elements:
{ Named Entity Recognition (NER), which is a module for
recognizing entities in bunches of text. A machine learning mechanism based on
OpenNLP 5, a set of language features and a set of annotated documents
for nding candidate NERs, enhanced by the use of dictionaries is used [
The high level data analysis of the system includes a number of components that
combine the outcomes of the deep data analysis and they include:
{ Discovering social media in uencers [
The system frontend consists of both web and mobile applications that utilize the data collected and analyzed in order to present reports, visualize data and make it easy to explore the combined information.
The web and mobile applications will have a public part that will make parts of the collected available to public. This is a news aggregation service including rich media format of data as well as interconnection of information and multilingual content. The same is for the mobile application which ca be formulated in order to enhance portability and usability of the presented content. 4
The design and development of the proposed system consists of a new and innovative product for the international market, which is expected to be the attraction for many companies and organizations primarily organizations that operate internationally. The absence of specialized competitive products in this eld o ers a signi cant advantage and allows it to be a leading player in the Greek market, which is the country of origin, and to penetrate the emerging and demanding international market of high-volume data analysis technology by providing innovative services and products.
All the aforementioned, is expected to provide a new dynamic in the eld of application development in the referred emerging sectors. This is achieved by using state-of-the-art technologies and methodologies together with the extensive knowledge in the eld by the partnership. At the same time, within the framework of the proposed project, the know-how acquired in the areas of large volume analysis is fully exploited, thereby enhancing the company's policy towards the increased use of cutting-edge technologies, as well as the partnerships' research background. Finally, we should consider the valuable know-how acquired by all the participating bodies during the implementation of the proposed project through the two research organizations, which will be done by the research and development in order to achieve the desired objectives. The know-how to be transferred will improve all the organizations' and especially the company's scienti c potential by increasing its knowledge and expertise and consequently the company's capabilities for future support as well as developing new applications and undertaking new research projects in the context of its activities.
Focusing on cultural related organizations it is possible to nd opportunities that these venues never had. It is important to note that cultural spaces have recognized the important role of technology and online synchronous and asynchronous communication, and are willing to utilize modern and edge-cutting technological features in order both to enhance the experience of the visitors as well as attract a broader audience. In this scope, it is extremely di cult for people related to arts and culture perform an advanced step towards analyzing the impact of their presence and marketing procedures on the internet. PaloAnalytics project can play the role of the companion when it comes to their online presence. The project can help recognize the supporters and fans, can measure the impact of the online marketing strategy, can keep a record of other spaces' impact or connection and can help towards the improvement of the online presence. 5
We presented the project PaloAnalytics, which is reaching its rst year of undergoing. During this period the rst crucial steps have been made, including the de nition of the system use-cases, the formulation of the system architecture, the set-up of the system infranstructure, as well as the design and initiation of the rst system components. Furthermore, the business-case is completed and the implementation of the rst sub-systems is almost nalized. The infrastructure of the system is set-up and the means of integration are de ned. An interesting feature of the project is the participation of two research laboratories from two di erent institutions in Greece, which will join their research teams to produce the results of the project. In order to achieve the objectives of the project, cutting-edge technology and algorithms are used, which means that the participants will join forces towards the research.
Despite the fact that the actual outcome of the project is minimal compared to the algorithmic procedures that lead to it, a number of related research elds will be explored during the design and implementation of the components. First of all, data mining algorithms will be researched in order to produce the optimal solution for fetching data. Furthermore, the infrastructure that stores the data is the basis of the system and as such its design and integration is part of a research and development procedure. On the other hand, a number of algorithms and techniques including deep machine learning will be investigation in order to achieve procedures listing: clustering of data (including text objects deriving from social media), summarization of clusters, named entity recognition, sentiment analysis, aspect mining and breaking news de nition. Furthermore, apart from the core algorithms, a number of \high level" procedures are required in order to achieve the complete set of project scopes. These include in uencers mining, semantic web, network analysis, campaign impact, swot analysis and more, which are based on the metadata that accompany the information collected and processed.
It should be noted, that all the aforementioned are not just part of a research procedure; meaning that the research should not stand on the feasibility and soundness of the results. The system is a production based environment targeting large business and organizations, which can even test and formulate the procedures and the use-case scenarios. It lies on the ground of applied research and it is expected that all the implemented solutions will be able to endure large volumes of data, users and demanding procedures.
As far as the role that the system can play for cultural organizations it is clearly de ned as an important one. Speci cally we de ned the system as a valuable companion that can totally alter the procedures of online marketing strategies and social media interactions. The system can be used to examine the behavior of the users towards exhibitions and presentations as well as towards individual cultural objects. The project can be the beginning of a new era in cultural informatics, acting as a novel pioneer procedure, that can involve edge cutting technologies directly on the relation of the organizations and visitors introducing a new way of mass culture.
This research has been co nanced by the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH CREATE INNOVATE (project code: T1EDK-03470)