How to avoid clutter in your search January 22, 2024 20:44 Updated Discover the necessary steps to reduce unwanted publications in your search. The Search Configuration is a crucial step in gathering and categorizing publications and interactions on networks, to be viewed and analyzed in other modules of the platform. A successful Search Configuration ensures the quality of collected data, which is subsequently structured with the assistance of STILINGUE Artificial Intelligence and displayed in customizable graphs, dashboards, and reports. During the configuration stages, it is common to encounter what we refer to as garbage, meaning data that does not align with the goal of your search and, therefore, is not relevant for collection. Some tips have been compiled to avoid this situation: 1. Conduct preliminary research on the subject using Radar and Twitter. Before including terms in your search, perform a search using the Radar tool and on Twitter (directly on the social media platform) to identify the key terms related to the topics appearing in the returned publications. This way, you'll be able to select what is not interesting to collect and can negate terms without relevance within the Groups. 2. Correctly use the operators In certain situations, you may encounter broad terms registered in your search, and you will likely collect publications outside your analysis scope. In this case, it's worth directing the collection to the context using operators that help delineate the limits of your search, such as the "context:" operator. Other operators like "exactly:", &&, among others, can be used to refine your search. 3. Negative Terms in Context If you notice terms that frequently appear and are unrelated to your research, it's possible to negate a term within the context to avoid collecting irrelevant publications for the analysis you want to perform. When negating the term in the context, the system will consider the entire search - in other words, the unwanted term will be disregarded in all collection groups and classification themes. 4. Caution when activating the Monitor searches in Context Button With this button activated, publications containing the terms registered in the context area will be collected - in other words, the context will have the same function as a group. These publications will not be automatically associated with the groups in your search, but they can be retrieved by applying a filter for "Unclassified Publications" in the Publication Filter. It is also important to note that this option may bring a higher volume of posts than expected (and consume space in your contracted plan) if a broad term is included in the Context, for example. 5. How to define the Context For the context of your research, think about terms you want to collect with your Groups, terms that relate to your analysis objective, and include the widest range of possible combinations: singular, plural, infinitive verbs, other conjugations, or spellings that may have more than one form (whether correct or not). Simultaneously, perform manual searches on social media and also use the Radar for real-time queries. You can input the terms from your Groups when searching directly on social media and observe the posts that are returned, checking if they align with your research. 6. How to register your brand It is common for brands to have names used in everyday life, and when registering this name, many irrelevant publications may be returned. In this case, some practices can help refine the search: Use the context operator linked to the brand name; Consider possible spelling errors; Use articles before the brand (Example: a STILINGUE, from STILINGUE, in STILINGUE). 7. Use the Overview to find unwanted terms Any irrelevant term can go viral and bring garbage to your search. In this case, it is ideal to always check in the Overview what are the main terms collected and if they are relevant to the context of your brand. If you find unnecessary terms, negate them in your Search Configuration. In the same area, it is also possible to check who the major influencers in your search are and whether they are aligned with your brand. 8. Banning users to prevent spam At times, it is possible to come across users who make numerous posts in a short period (such as bots) that are not relevant to the researched universe but have had their posts collected due to the way the Search Configuration was set up. In such cases, there is the option to ban the author of the post. Once banned, the platform will no longer collect anything about this particular author, whether in Listening or any other module of STILINGUE. All posts collected from the banned user will be deleted. Note: When banning a user from STILINGUE, the posts from that user are removed from the client's search. However, when the banned user mentions the proprietary page, comments made in this mention of the user continue to be collected because, despite the mention being deleted due to the ban, it remains in the collection queue for 30 days, resulting in the collection of comments in that mention. The collection of comments from this mention that the banned user made continues for up to 30 days after the mention (the collection rule we have today). Therefore, in cases where clients aim not to collect comments in mention, this workaround of banning the user does not work. 9. Remember: Search Configuration is a living organism Always be aware of what the Groups are collecting and how the Themes are categorizing these posts. Evaluate the collected data frequently: it's important to check, at least once a week, how your search is running to identify, during collection, whether there is a lot of junk or if the presented data is within - or close to - the research objectives. This way, you will be able to negate/include terms, ban users, and steer clear of undesirable data. For more information, visit the discussion on the subject at our community or videos on our channel. 😃 Related articles Technical Data & Other Definitions - STILINGUE Create New Alert Collection settings and status Creating interactive messages in WhatsApp Automation | Boolean Operators