Open data - Download the Knowledge base
You are free to download the data of this Knowledge base.
To do this you must be an authenticated user: log in or sign in now.
All the data are licensed as Creative Common CC-BY 4.0.
Text and opinion mining for policy making
This use case covers the methods that can assist policymakers throughout all stages of the policy cycle. It explains the sources for these data and how the outputs can be used to gain. understanding of stakeholders’ and citizen’s opinions on policies and strategies
Type of content:
Assets
Type of asset:
Use case
Big data potential
Yes
Policy domains:
Agriculture, Fisheries, Forestry & Foods
Open license availability
No
Addresses:
| SWOT Analysis for Text and opinion mining for policy making |
||
|---|---|---|
| Helpful | Harmful | |
Internal |
Strengths• Assist policymakers throughout all stages of the policy cycle: from getting feedback on different policies, creating a map of a current crisis, or shedding light on places where citizen’s feedback is needed. • These methods are tools that can be used across the board on many policy areas and topics. • Text mining allows finding trends in a large canon of text. The method assists in highlighting topics by creating numeric indices. It can create summaries of the frequency of a word, clusters of words, trends, and the like • Opinion mining refers to analysing positive or negative valences around topics: Opinion mining allows for the categorization of content to either binary values of positive and negative or scales of values such as very good, good, satisfactory etc. This is done through algorithms that classify documents and search key words that underline a sentiment. • Text and opinion mining are based on natural language processing, a computational process that retrieves high quality information from texts by detecting of patterns and trends in a corpus. • The tools offer visualisation of the data that can help the policymaker understand the complex data. |
Weaknesses• Text and sentiment analysis are strong only in the main European languages • Sample of the data and data analysis abilities of both text and opinion mining can entail a bias towards specific populations or types of stakeholders. populations. • Difficulties in terms of understanding the cultural context of a sentence, detection of sarcasm, and typos • Data ownership: Many sources on the web are prohibited from re-use of the data and some of them might contain private information (need to check that the sources can be used for these types of analyses). In many cases, the raw, existing data that is mined continues to be owned by the original authors and platforms, while the results of the text and opinion mining analysis can be owned by the policymaker that funds and oversees this process. -Data privacy: anonymisation and aggregation of data are among the points of attention in publishing the results. • Legal and cultural aspects of freedom of speech |
External |
Opportunities• Can be undertaken on many types of text from different types of media. Examples are social media, online and offline newsletters and study reports, letters, blogs and other documents by experts and citizens. • Many research projects, including EU funded projects, are operating to improve text and opinion mining for economic and research purposes. • Collecting data for framing policy: by using text and opinion mining on social media networks, policymakers can gather information that can allow them to understand the stakeholders’ needs and wants for a societal issues and an upcoming policy. • Creating a map of the current state of opinion and satisfaction levels from different groups of stakeholders: during the implementation stage of policies, opinion mining can help detect the satisfaction level from the policy interventions launched or the policies that have been adapted. This can inform policymakers in discussions about further improvements of policies. • Evaluating the implementation of policies: halfway or at the final stage of the policy cycle, text and opinion mining can help summarise the feedback of stakeholders and to feed it again into the (re)design of policy interventions |
Threats• Data source, Quality of data: Analysis of social media for example, can exclude some populations, like the elderly or lower class, from participating in the process, since they are not using social media networks (or using them to a lesser extent or for different purposes). Surveys can help in actively seeking feedback however, there can be gaps and inconsistencies in the data since people tend not to answer full length answers in surveys or have incomplete answers that can burden the algorithms. (policymakers should be clear about the population they want to include or want to hear from). • Data ownership: many sources on the web are prohibited from re-use of the data and some of them might contain private information. need to check that the sources can be used for these types of analyses). In many cases, the raw, existing data that is mined continues to be owned by the original authors and platforms, while the results of the text and opinion mining analysis can be owned by the policymaker that funds and oversees this process. -Data privacy: anonymisation and aggregation of data are among the points of attention in publishing the results. • Legal and cultural aspects of freedom of speech |
