Before the internet emerged and the web was created, most of the communication was carried out orally, in which there was no proof or evidence of these conversations, however currently this is very different than anything else due to the internet revolution.
Computer systems have been created that are represented by a large number of tools and processes through which people can learn faster and better.
Without a doubt, it is a challenge to discover knowledge in large volumes of data that is generated in conversations. For this reason, text mining emerged, and represents the newest research area of automatic text processing
Before the emergence of the social web, people made recommendations to each other regardless of the field, however in our days the internet makes it possible for users to find opinions from all areas and people from all over the world.
In the aspect of companies dedicated to the manufacture of products and services, analyzing the opinions of the same that circulate on the Internet becomes a vitally important strategy today.
In this way, companies are aware that there is a very close relationship between the investment made in optimizing the brand's social climate and the performance of the products they manufacture.
Mining of data, texts, feelings
Before the internet emerged and the web was created, most of the communication was carried out orally, in which there was no proof or evidence of these conversations, however currently this is very different than anything else due to the internet revolution.
Currently, the data of the conversations is stored at an accelerated rate, people and organizations participate in exchanges of information and this is done through emails, virtual chats, text messages, blogs and forums.
The advances that have been made in natural language processing represent opportunities for documents to be analyzed, and the web allows areas such as schools to present significant advances in a short time.
For this reason, computer systems have been created that are represented by a large number of tools and processes through which people can learn faster and better.
Data mining refers to the process of discovering knowledge through the analysis of large volumes of data.
The main applications of education-related data mining techniques are:
- Personalization systems: through these, as much data as possible can be obtained from users with the aim of improving the service that is being provided to them. Recommended systems: their usefulness is manifested in offering the user different topics that may be of interest, this is done by comparing profiles of these with characteristics of the topics. Modification systems: refers to the detection of needs by making necessary modifications to improve the user experience. Irregularity detection systems: these are based on the error detection.
Text mining
Text mining is a specialized discipline in obtaining information that is not explicitly defined through identifying patterns in a text.
Undoubtedly, it is a challenge to discover knowledge in large volumes of data that is generated in conversations. For this reason, text mining emerged, and represents the newest research area of automatic word processing.
Text mining, “is defined as that automatic process of discovering interesting patterns in a collection of texts, these patterns must not be explicitly in any text in the collection, and must arise from relating the content of several of them” (Barrio, 2014)
For example, on Twitter there is until it could apparently be an indicator of topics of interest, however thanks to data mining users can be classified based on their preferences, ideas and opinions. (semanticwebbuilder, 2016).
The data mining process consists of the following stages:
- Pre-processing: at this stage the texts are transformed into some form of structured or semi-structured representation that somehow makes subsequent analysis easier. Discovery: at this stage the intermediate representations are analyzed with the main objective of establishing patterns of interest or new knowledge.
Text mining application
Data mining applications include the following:
- Information extraction: there are a large number of documents found on the network, text mining allows the extraction of relevant information from large amounts of texts and manages to define and relate entities and thus reveal relevant information
of the texts, managing to define entities and their relationships, revealing important semantic information which is used as metadata, which at the same time is added to web content and thus facilitates the understanding of these documents.
- Sentiment analysis or opinion mining: it is the analysis of the opinions generated by social network users and helps to reveal information of importance for specific topics and in this way it can support different fields such as business intelligence and decision-making. consumer decisions. Document classification: this is a key aspect to facilitate the retrieval and navigation of documents since if you want to obtain information from different areas of project development it would be very difficult, however thanks to the use of mining algorithms Documents can be grouped data, in order to obtain information that describes each group to make it easier to understand each one of them.Its objective is to obtain general descriptions of a set of documents that belong to a specific topic. The methods can be classified into two categories:
- Extractive summarization: summaries that are made up of information units extracted from texts Abstract summarization, in this the synthesized information is not necessarily formed by information units contained in the texts.
Challenges of text mining
Text mining must face the following challenges:
- You must know the context in which the contents of the different tasks of text mining were generated, since you must know who the author is, the region in which it is located and the historical moment in order to understand the information. beyond obtaining information and seeking to obtain knowledge, this implies the transformation of the information extracted in different tasks in the area into a language that can be understood by the machines. It is necessary to analyze very large amounts of information, this can be very difficult when it comes to efficient processes and parallel algorithms must be defined to take better advantage of the current computing infrastructure. (Lozada, 2016).
Mining feelings
Before the emergence of the social web, people made recommendations to each other regardless of the field, however in our days the internet makes it possible for users to find opinions from all areas and people from all over the world.
In the aspect of companies dedicated to the manufacture of products and services, analyzing the opinions of the same that circulate on the Internet becomes a vitally important strategy today.
In this way, companies are aware that there is a very close relationship between the investment made in optimizing the brand's social climate and the performance of the products they manufacture.
Analyzing feelings in opinions is a procedure that is carried out applying data mining techniques on the texts that express them.
Data mining involves the extraction for a specific purpose of the information that is implicit in the data, this information is previously unknown and may be useful for certain processes.
In this way, it can be understood that data mining prepares, probes and explores the data to extract information that is hidden in it.
Data mining encompasses all the techniques aimed at extracting actionable knowledge that is implicit in databases.
The bases of data mining belong to artificial intelligence and the models extracted using data mining processes address the solution to prediction, classification and segmentation problems.
Goals of sentiment mining
The mining of feelings has the main objective of creating web tools that make it easier to extract the opinions of different social networks to later be able to classify them.
However, this tool must be able to provide an interface that is entertaining and friendly for system training that specializes in the classification of opinions, and allows showing results in a more comfortable way to interpret than if they were simple data.
In this way, it will be possible to demonstrate the importance of user opinions and how these influence the popularity and integration in the market or also in the failure of certain products. (Stripped, 2011)
Some history of mining feelings / opinions
The term opinion mining first appeared in an article that was published at the www conference in 2003, due to this publication the term is very popular in communities strongly associated with web search or information retrieval.
The article states that the ideal opinion tool should process search results that are carried out on the established topic.
However, the term sentiment analysis has suffered something similar appeared in 2001 in articles related to marketing and took relevance in the world of computers when it was published in subsequent works in these areas. (Lopez, 2011)
conclusion
Data, text and sentiment mining is of great importance in today's world since thanks to this, the processes of obtaining and decoding information that exist on the Internet and that contain relevant data for data collection can be synthesized decisions of many companies, as well as to improve customer service and product development processes.
Data, text and sentiment mining facilitates the activities of all people around the world and aims to continue advancing by leaps and bounds.
The biggest challenge in mining is to obtain accurate data that is reliable and classified according to its origin in order to achieve the best results for the best decision-making of humanity.
Bibliographic references
Barrio, E. (2014). Text mining for the analysis of the collaboration in SIENA. University of the Laguna.
Lopez, L. (2011). Classifier based on sentiment analysis. IUMA.
Lozada, A. (2016). semanticwebbuilder. Obtained from semanticwebbuilder.:
www.semanticwebbuilder.org.mx/es_mx/swb/Mineria_de_textos_y_sus_aplicaciones
Pelado, J. (2011). Data mining applied to sentiment analysis in internet opinions. University School of Informatics.
Semanticwebbuilder. (2016). semanticwebbuilder. Obtained from semanticwebbuilder:
wb / Processing_of_Natural_Language_and_Mineria_de_textos
Thesis proposal
Opinion mining and its current impact on the development of multinational companies.
world Wide Web
Download the original file
