FAQs - Search Configuration – Blip | Blip Help

Index:

When registering a Facebook operator to collect posts from an ID, the search didn't collect the comments. What might have happened?

If the operator (facebookcommentsid, facebookcommentspid) is not working as expected, it is necessary to check the registered ID number. In the case of posts with videos, the registered ID should pertain to the post and not the video.

When adding a URL in the Add Blogs section, can I ensure that 100% of the posts from that page will be collected?

Even after adding a URL to prioritize the collection in your search, this action does not guarantee the completeness of post delivery, as blog and website collection depends on an external aggregator.

Additionally, the content on the page must align with the terms previously registered in the Groups of your Search Configuration.

If I remove a Group, will the posts from that Group also be removed?

It depends on how you save your search. If you choose "Save and reprocess," all configuration changes are considered, and the system adjusts the last 100,000 posts to the new search configuration. Therefore, this reprocessing would recognize that the configuration no longer includes a Group and would remove the data that no longer fits this configuration. If you opt for "Save without reprocessing," the posts are not deleted, as this feature was designed to only change the posts that will enter the search without modifying what was collected previously.

How does language identification work for posts that don't contain text and are composed of emojis, links, numbers, or user mentions?

Depending on the publication, language identification through text is limited by factors such as content consisting solely of links, emojis, user mentions, numbers, email, and an insufficient amount of letters. In these cases, language identification will be carried out using language information available in links, mentions, or hashtags. For example, a post containing only the link www.stilingue.com.br or the hashtag #StilingueBrasil will be classified as Portuguese, while a post with only the link www.stilingue.com/en-us/ will be classified as English.

In the absence of this information, language will be classified through metadata, which means data related to the main publication, such as the location and user biography on a social media platform. Currently, this solution can be applied to posts on the following social networks: Twitter, Facebook, Instagram, and Youtube, with location metadata available only for Facebook and Twitter.

In the case of posts that lack any type of classifiable metadata, the language will be classified as "Other." This classification can be filtered using the Listening Post Filter under WHERE > Language.

Will Proprietary Connections collect everything, or only what is related to the search configuration?

Proprietary Connections collect all posts related to the page. For retroactive searches, the last 100 posts from the last 7 days will be collected for Twitter, and the last 100 posts from the last 30 days will be collected for Facebook, Instagram, LinkedIn, Google Play, and Youtube.

Is it necessary to renew all active hashtags every 7 days?

No. The active hashtags will be automatically renewed while they are in the "active" box. The 7-day waiting interval is a norm of the Instagram API and occurs only when a hashtag is paused.

Is there a limit on the volume of posts for hashtag collection?

Yes. Collection is done through a request made every 15 minutes, returning 50 posts at once. If a particular hashtag has more than 50 posts, there may be loss of collection every 15 minutes. This is a guideline from the Instagram API.

To identify the exact time of an Instagram post, access the post on the social network and right-click on the date located below the likes. Then, click on the "Inspect" option as shown in the GIF below:

A side screen will open with details of the post's code. Note that the highlighted section in the code “datetime=” precisely indicates the date and time the post was made.

Another important point is that the time is shown in Greenwich Mean Time (GMT). To calculate the time according to the Brazil time zone, it is necessary to subtract 3 hours from the value shown in the code.

I've negated a term in one of my Groups, but they continue to be collected. What might be happening?

It's possible that this negated term may have been added to only one Group, which means that only that Group does not collect posts with the term. To "block" the entry of posts throughout the search, it's necessary to insert the negation operator in the Context.

Is there a limit on the number of operators (terms) per group or search?

There is no limit to the number of terms, you can check how many terms were entered in the Groups and Themes fields in the Research Summary:

It's important to remember that as new terms are added to the Configuration, the research collection time may increase.

Is there a limit on the number of operators for groups or themes?

There is no limit, but rather an indication of the quantity of terms per plan. This is because, as more words are included in the Configuration, the time on the clock indicating the collection cycle increases. If the time reaches 24 hours, there may be data loss. In that case, it is necessary to remove some terms from the search configuration.

In the agency where I work, everyone has admin access to the pages. Will each person be entitled to collect 30 hashtags?

No. The total number of hashtags is not defined per user. If the Meta business account has more than one page, the set of pages has a total of 30 hashtags. In other words, the pages together have this limit.

Does Instagram differentiate between accents, cedilla, and plural forms?

Yes, what doesn't make a difference is the capitalization. If the hashtag you want to monitor has accents or cedillas, you'll need to include all possible forms of writing and even plural options, which will impact the available limit. If you want to search for a specific hashtag, you can use the exact operator.

Example: #child, #child, #children, #children.

To check if Instagram distinguishes these variations, simply perform a manual search on Instagram itself and see if the number of available posts for each option is different.

What happens if I only put the hashtag in the group? And if I only put it in the hashtag manager? Will the search collect it?

You need to have all the hashtags set up in the manager, as it allows STILINGUE to go to Instagram and collect the posts related to these hashtags. However, this isolated action alone does not guarantee the inclusion of the posts in your research.

What will ensure the collection are the terms contained in the groups, so if you want to collect any and all posts with a certain hashtag, you need to include it in some group as well. An important point to emphasize is that it is not necessary to include the "#" symbol in front of the terms, whether for those inserted in the manager or in the groups.

However, if you only want to collect posts with these hashtags when some other term from your groups is mentioned, just put the hashtags only in the manager. By doing so, posts originating from these hashtags will be available for collection and, if they have other corresponding terms in the Search Configuration, they will be collected in the groups.

What happens if I'm using my 30 hashtags, and on the day I want to launch a new campaign, I want to change them?

According to Instagram API guidelines, it will only be possible to include a new hashtag after a 7-day pause.

Why was a Facebook post with a term included in my Search Configuration not collected but appears in Radar?

In some cases, a post that is not in the Listening database may appear in the Radar because the collection was done through the Influencer Hub. Another point to consider is that the collection of Facebook posts is not guaranteed solely by the inclusion of terms. It is necessary to include a network operator to monitor the page responsible for the post, to ensure that comments will be added to the search.

Can I use the same hashtag in different panels without counting towards my limit?

Yes, if these panels are set up under the same Stilingue account.

When my search reaches the limit of the contracted plan, what should I do?

The account's master will receive an email alerting that this limit is about to be reached. If the account has more than one master, all of them will receive the email.

There are several options to make the configuration lighter and thus normalize data collection:

Review the search configuration (removing excessive or irrelevant terms);
Deactivate unused channels;
Enable the anti-spam feature;
Manually delete non-relevant posts directly on your wall, under "view advanced options."

If you'd like to keep your Search Configuration as is and want to upgrade your plan, please get in touch with your Success Manager.

If my page experiences authentication failure, can it affect the hashtags registered in the search?

Yes. When authentication failure has been active for 7 days, the hashtags entered in the search will be removed from the Hashtag Manager and cannot be retrieved. This removal of hashtags is done by the social network itself because they are deemed unused. It's important to remember that "space" for new hashtags becomes available after 7 days without collecting posts. In other words, after this period, hashtags that haven't been monitored are deleted from the Management box. For Instagram, it's the same behavior as if they were not being used for collection, which is why they are automatically eliminated.

Will a post that has its text edited with an Instagram hashtag registered in my search be collected by the platform?

STILINGUE does not guarantee the collection of posts that have had their text edited. In the case of Instagram, when you hover over the post's text, you will see if the original content has been edited. When the text of the post is edited, it is not possible to ensure its collection because there are no traces of the content of the post prior to the edit. Therefore, it is quite likely that the original text did not contain the term registered in the search, in this case, the hashtag.

Can news collection (portals) also be delayed in the search?

The delay in collecting articles and news is usually associated with publications that are added to the search through STILINGUE's clipping service. The clipping service delivers offline press publications when the terms in these publications are related to the Search Configuration, they are collected. Some news articles have a publication date prior to the date when the news was made available through the API, which STILINGUE uses to carry out the rest of the collections.

Will the collection of a monitored Instagram Business Discovery profile have the metrics of each post updated?

Yes, the metrics must always be displayed with the latest updates.

Currently, we have 3 sources for collecting Instagram posts:

Via hashtags;
Via business discovery (profiles set up in Edit Search);
Via business discovery (profiles existing in the Influencer Center database).

Only the users retrieved from posts in the third category (profiles existing in the Influencer Center database) are added to the user and their posts' metrics update queue. It is worth noting that the collection of posts from profiles registered in Edit Search does not add the posts to the metrics update queue due to compliance issues, as the same metrics update queue is used for all clients. If a post is added, it can be added to all searches as long as it matches the configuration, which cannot happen in this collection, which must be restricted to the account in which the profile was registered.

Is the Forum channel collection (existing in the Search Configuration) active in the system?

Currently, this data collection is only active for Yahoo Answers. Collection for this channel will be restored in the system soon. As per the tool's definition, the author's data appears as "anonymous user," and card metrics are not sent. However, the text appears normally within the Publication Wall.

Does the Instagram Business Discovery collection have any relation to proprietary pages?

Yes, the collection is done based on the user's token, meaning each registered proprietary page in STILINGUE enables the monitoring of up to 30 business or creator profiles on Instagram.

When registering Instagram profiles in my Search Configuration to be monitored via Instagram Business Discovery, will all posts from each profile be collected?

Similar to other social networks in the "Add Profiles to Search" section, when registering an Instagram account, the system will monitor the posts from that profile. However, it will only collect those posts that mention any of the terms registered in the groups of the Search Configuration.

Nonetheless, it is possible to configure the monitoring so that every post made by the page is collected by the search without the need for a match with the Search Configuration. Follow these steps:

Click on the registered page, then in the "Sort into:" box, select the group that will gather all the posts from the profile.
Upon selecting the group, an Instagram network operator will be created in the chosen group.

It is also possible to select a theme that will gather all the collected posts from this page. To do this, just repeat the process, but instead of a group, choose a theme. When selecting the theme, an Instagram network operator will also be created within the chosen theme.

Currently, we have 3 sources for collecting Instagram posts:

Via hashtags;
Via business discovery (profiles set up in Edit Search);
Via business discovery (profiles existing in the Influencer Center database).

Only users retrieved from posts in the third topic (profiles existing in the Central Influencers database) are added to the user and their posts' metrics update queue.

It is worth mentioning that the collection of posts from profiles registered in the Edit Research does not add the posts to the metrics update queue due to compliance issues, since the same metrics update queue is used for collecting all clients. If a post is added, it can be included in all searches as long as it matches the configuration, which cannot happen in this collection that must be restricted to the account in which the profile was registered.

Important! To ensure the page's inclusion in Listening, it is necessary to use the ID descriptor within the Group. It is not required to be in the profiles, as they serve different functions.

Remember:

Add Profiles: it enters the search only if there is a match with any descriptor of the groups;
Network Operator: collects all profile's publications. When inserted in a group, it collects everything it covers and ensures the match with the specified group.

When adding a brand profile from the suggested Groups, is it also necessary to include the same profile in the Add Profiles area?

In some cases, yes. The general rule is: when you use the suggested Groups, there is an option to add profiles or descriptors of a particular brand. If the desired profile is added to the group, there is no need to include it in "Add profiles" as well. However, there are two situations where inclusion is necessary: for "regional" or "smaller" pages, it is important to insert them in the "Add Profiles" field to ensure data collection. The other point to remember is that, in the case of Twitter, the collection of posts is sampling (except for Twitter Enterprise).

If I include a term with diacritics in my group, will it be collected in that way?

First, it's important to understand that the Search Configuration groups do not differentiate accents and other diacritical marks, such as cedillas, in the entered term, regardless of the selected language (Portuguese, Spanish, or English). This means that when entering the term "Pará," the system will search for both the state of Pará and the preposition "para," which does not have an accent.

However, there is a way to ensure that the specific term with accents is collected in your group.

When adding a profile in the Add Profiles area, do I need to include operators in a Group or a Theme as well?

If you wish to collect posts from this profile within a specific group, yes. If only the profile is placed in this area, the collected posts will be included in the search as unclassified (meaning they won't appear within any Group or Theme).

It's also important to highlight that posts from the inserted profile will be collected only if they are related to the terms included in the Groups of your research.

Enabling the Monitor Context Searches option will collect any Context post?

Yes. And this is why it should be used with great care not to exceed the data collection limit in this monitoring, which can be very extensive.

The collection carried out by Context can be checked by disregarding all filter groups and, thus, filtering for unclassified data in Groups.

I registered terms in other languages in my search, but they were not collected. Why did this happen?

STILINGUE's Artificial Intelligence (AI) employs a language classification methodology capable of understanding when a publication is mixed (i.e., written in multiple languages). To detect and display words in other languages in the results, it's necessary to define the languages that will be searched for in the initial configuration stages.

First, define the Groups for your research:

When defining the Groups for the research, it is also necessary to specify which languages will be collected, if there is more than one.

After that, define the themes:

When defining the Themes that organize the collected posts, it is also important to select the languages, if there is more than one.

How does STILINGUE perform sentiment classification when a user interacts only with laughter?

In the case of posts that contain only laughter-indicating language, such as: kkkkk, hehehehe, hahahaha, hehehe, and others, the content undergoes a process called normalization so that all variations of laughter are understood in the same way.

Although normalization makes it easier to identify a text with laughter, it is important to emphasize that this type of interaction is not considered by the STILINGUE Artificial Intelligence engine when performing sentiment analysis. In other words, our AI will not automatically classify a message with laughter as positive, neutral, or negative.

This is because a laugh cannot be automatically classified as a positive sentiment, as it is often used in ironic or neutral and negative contexts. Therefore, the AI cannot determine the polarity of a post with laughter if there is no other textual element providing context for that post.

How is sentiment classification done for proprietary posts when the "Sentiment Analysis on Proprietary Interactions" option is disabled?

If you prefer automatic sentiment classification not to be applied to posts and comments made by the owning pages, you can disable this option. When this option is unchecked in the "Settings and Collection Status" area, only owning posts and comments will be marked as "Unclassified," without the polarity of Positive, Neutral, or Negative. It's important to note that this function remains disabled unless a sentiment classification rule has been configured previously, as this rule will annotate the owning post with the corresponding sentiment.

It's also worth noting that this option only applies to owning interactions. Mentions, comments from other users, and visitor posts continue to be classified according to the polarity of each post.

How can I find the ID of a YouTube channel with a customized URL?

Observing the YouTube URL, it is possible to identify the channel or video ID. The sequence of letters that appears in the link after the segment "/channel/" corresponds to this ID, as shown in the image below:

This action is necessary so that YouTube network operators can be inserted into the Search Configuration.

When the URL is customized (that is, the channel name is inserted in the link), the ID is not displayed. In this case, you can find the ID by following the following steps:

1. Go to the homepage of a channel on Youtube, which contains this type of URL:

2. Click on any video from the channel, so that you are directed to the page where the video is located:

3. Then, return to the channel's home page by clicking on the area located below the video that displays the channel's name and avatar. The channel's URL will be presented with the ID:

How does language classification work for posts that contain more than one language?

Currently, STILINGUE uses a language classification methodology that can understand if a post is mixed (i.e., written in more than one language) and then detect the languages. In addition to the detection done by metadata, the system will also consider the languages selected in the collection groups created in the Search Configuration during the detection process.

Then, the mixed publication will be classified based on the language that has the highest weight in the system's assessment. The classified language can be viewed through the publication card (found in the Listening Mural, Proprietary Metrics, and other areas of the platform), below the related image.

This methodology will not be applied in the following situations:

Publications that contain names of series/movies in another language. In this case, only the main language of the message will be considered, not that of the mentioned film/series. This process is carried out with the assistance of a specific internal dictionary, so that these terms are disregarded.
Example: the publication "I love watching How I Met Your Mother" would be classified with the Portuguese language, and the name of the series would be disregarded in the analysis.
Publications made in a primary language but containing hashtags in another language. In this context, the hashtag term will be disregarded to prioritize the primary language of the publication.
Example: the publication "#tbt today in tribute to the 27 years of #ufc When I visited the museum [...] #brazilianjiujitsu #fit #fitness #ufc #bjj #workout #sport #gym #fight #boxing #judo #knockout #mixedmartialarts #meme #memefight #ilhadaluta" would have the English hashtags disregarded in the classification, so the primary language would be Portuguese.

How does sentiment classification work for terms written as acronyms or abbreviations in posts collected by STILINGUE? Example: tbt, sqn, bjos, amg, ctz...

When a term written as an abbreviation of a word or an internet acronym is collected by the platform, a normalization process is performed, which returns this abbreviation to its original form. After the normalization step, sentiment classification is carried out based on the context presented in the publication. This normalization is one of the internal text processing steps and is not visible in the interface.

Example:

Text collected: "amg, eu amei isso, sqn"
Processed text: "amiga, eu amei isso, só que não."
Text that appears on the platform: "amg, eu amei isso, sqn"

This normalization is done by the Machine Teaching team based on an extensive and constantly updated dictionary of terms. The platform's interface remains unchanged: the replacement of the abbreviated term with the full version is done automatically, as well as sentiment classification.

In the case of the abbreviation "sqn," which is ironic, it is classified as Negative on the platform, depending on the context of the publication.

Is it possible to choose a color for my search?

Currently, it is no longer possible to choose a color to customize your board. When creating a new survey, the card will have the default color, which is green. However, if it is paused, the card's color will automatically change to yellow, and if the collection limit is reached, the card will change to pink.

Note! It is important to remember that existing surveys will continue with the color selected at the time of creation, but after that, they cannot be customized.

Is it possible to monitor the same hashtag on another platform?

Yes. In this case, the hashtag will appear in red within STILINGUE's Hashtag Manager, but since monitoring continues to be active on the other platform, the hashtag will not disappear after 7 days. The same logic applies to another account that has the same connected page or for another search within the same STILINGUE account.

To assist in verifying which tool monitors the same hashtags as STILINGUE, you can access the Facebook settings for your proprietary page. By clicking on:

Settings and Privacy
Settings
Apps and Websites

You will find all the apps and websites listed, including the tool that monitors the same hashtags.
Attention! All users who have permission to manage the account (administrators) need to verify the listed applications.

Is it possible to pause data collection/search?

Yes. On the War-Room's home page, select the account, hover over the search you wish to pause, and select the "Pause/Resume Search" button. To resume, follow the same procedure.

Is it possible for a post not to be collected by my search due to language detection?

Yes. STILINGUE uses a methodology for language classification that is capable of understanding the characteristics of a publication, and this methodology is built using Artificial Intelligence (AI) systems. In AI systems, there is not a 100% accuracy level, and a certain amount of acceptable errors is natural. Error rates are present in any system that relies on AI, whether it is within or outside STILINGUE.

The practices adopted to evaluate these classification errors are as follows:

For texts that have characters very distinct from the Portuguese language, very short texts with little context, this identification becomes a bit more challenging. For longer texts, several specific classifiers are employed. One of the practices used to make this type of identification is to separate the objectives:

High Coverage

Having high coverage with respect to the identification of defined languages in the research means collecting posts even if they are not in those languages.

High Accuracy

Being more restrictive in the collection, gathering only posts in the languages defined in the research. This means, in some cases, failing to collect posts that should have been included.

There is a balance between "High Coverage" and "High Accuracy" when classifying languages. Therefore, some posts may be excluded based on the classified language. This is not a significant number, as the priority is to deliver "High Accuracy" in the language classification for the research.

Is it possible to undo a search reprocessing?

It is not possible to reverse a reprocessing because the old state of the data is not retained by STILINGUE, due to the lack of historical data retention for changes caused by reprocessing.

Is it possible to use a comma in the Search Configuration?

No. Our platform does not allow the use of a comma or semicolon in the Group and Theme fields of the Search Configuration.

Is it possible to use the same owner's Instagram Business account to monitor different searches?

Yes. If you have more than one search panel, it is possible to connect the same account to different searches and, in each one, monitor up to 30 business profiles or creator profiles. For example, if the STILINGUE page is connected in three different searches from the same account, it will be possible to monitor 90 different Instagram pages, with 30 pages in each search.

How long does it take for the collection of posts from the profile monitored by Instagram Business Discovery to occur?

After adding a business profile in the Search Configuration that has posts made in the last 30 days, the data collection should take a maximum of 15 minutes to start.

Are emojis considered in the sentiment classification of posts?

Yes. Emojis determine the sentiment applied to the post, just like words. STILINGUE's library contains over 1000 emojis, the result of mapping based on users' social media posts. Only these mapped emojis are considered in the sentiment classification.

The library used for this sentiment analysis categorizes each emoji according to its polarity, which can be Negative, Positive, or Neutral.

These values are part of an equation that determines the sentiment of the publication based on the final result, just as it is done with the sentiment word dictionary terms.

The emoji's polarity is determined by considering its frequency of use on social networks. If an emoji is not used frequently, it is assigned a Neutral polarity. If this changes over time, and the emoji becomes more commonly used in a negative context by users, its polarity will shift to Negative.

Attention! If the publication does not contain text and consists only of emojis, the sentiment classification is based on the polarization of emojis to determine the sentiment of the publication.

Can I prevent a post from entering my search?

Yes. It is possible to blacklist terms that appear in this publication (and may attract similar posts) Additionally, it's also possible to block the user who made the post - in this case, all of their posts would also be blocked - or block an individual post, through the advanced options at the end of the publication card on the Publication Wall.

Is there any prioritization or difference in the collection time for Instagram Business Discovery that can vary depending on the contracted plans?

No. The collection should work the same way for all plans.

Are there characters that cannot be written in the Search Configuration?

Yes. You cannot enter the forward slash (/) and grave accent (`) characters in the fields of your Search Configuration.

In Instagram Business Discovery, when a monitored profile is deleted, are the collected posts also deleted?

No. When deleting a registered profile from monitoring, it's important to check if the posts are indeed no longer being collected. The time for the collection to finish after the profile is deleted is 15 minutes, so monitoring should no longer occur after that period.

However, deleting the profile will not remove the posts already collected. Therefore, if there is a need to no longer view the posts collected by the deleted profile in the search, you need to remove them manually.

Are exclusive news for subscribers collected by STILINGUE?

The collection of posts exclusive to subscribers is not guaranteed by STILINGUE. If this type of post is included in the search, the collection will be done through clipping. Clipping involves monitoring mentions of a specific brand or online and offline media outlet.

Does Instagram Business Discovery have retroactive collection of posts?

Retroactive collection has a limit of 100 posts made in the last 30 days.

What happens if I use more operators than allowed in the group?

The collection of the excess operators will not be counted within the search. This is because if you include more operators than allowed (per plan), this action will be blocked by the system.

What happens if I edit my Search Configuration but do not save the changes immediately?

There are two different scenarios identified for this action in the system:

1. You are editing terms in the Search Configuration, save your changes, and remain on the same page for some time. If you switch tabs, and in the meantime, another team member makes edits to the same search, when you return to this page, you will receive a pop-up notification that the page will be refreshed to provide you with the most up-to-date version of the search.

For this scenario, since you saved the changes, your edits will not be lost.

2. You are editing terms in the Search Configuration, but do not save the changes and switch tabs. If while you were away, another team member made edits to the same search, upon returning to the page, you will receive a pop-up notification that the page will be refreshed so you can access the most recent version of the search.

For this scenario, since you did not save the changes, the edits you made will be lost. You will need to redo the edits and save the search.

Therefore, it is important to save the latest updates in your dashboard to avoid losing any records.

What characterizes spam?

When the same text is repeated in more than five (5) different posts in the same search, it is characterized as "spam."

The anti-spam works in the same way as in email, blocking these posts from the search. This way, the data analysis will not be affected by an excess of repeated posts.

What are "invalid pages" for Instagram Business Discovery, and how can you determine if a page has been invalidated for collection?

There are some changes that can invalidate the pages registered for monitoring, and to resume the collection of posts, some actions need to be taken. Check out some of the changes in Instagram pages that can impact your monitoring and know what to do in each case:

Username Change: In this case, to resume monitoring, you will need to re-register the page in the Search Configuration and update the Instagram network operators using the new username.
Business Profile Status Change: In this case, it will no longer be possible to monitor it on the platform until the profile returns to being a business page.
Page Removal: If the page is removed from Instagram, it will no longer be possible to monitor it on the platform.
Connection of Monitored Profile in My Connections: If a page that is being monitored via Business Discovery is connected to the same search as an owner account, your registration in the Search Configuration will be automatically invalidated. Data from owner pages can only be accessed through the Listening area via Conversation Filter.

You will be notified by email that the page has been invalidated, and you can also check in the "Add Profiles" field in your Search Configuration. The concerned profile will appear in red:

To initiate the collection of a Business profile, is it sufficient to register it in the Instagram profile monitoring area found in the Search Configuration?

No. To ensure the monitoring of a registered profile, it is also necessary that the username of this profile is registered in a group and that the Instagram channel is active in the Search Configuration.

The combination of these three points will ensure the proper flow of data collection:

Profile registration in the monitoring area
Username of the registered profile in a group
Active Instagram channel in the Search Configuration

Why did my posts disappear after removing them from the group?

When a group is removed from a post, it will still be included in the search but won't be found by conventional filters, as the search takes into account the groups. To find it, you will need to search for uncategorized groups, meaning not selecting any group in the Post Filter.

Why are news articles collected by STILINGUE categorized as blogs?

STILINGUE uses a tool that monitors the feed of various internet sites. It's this process that extracts the posts that enter the Blogs channel. Among the feeds monitored by this tool are also news portals, and these posts end up being included as blog posts.

Why may blog posts take longer to appear in the search?

STILINGUE monitors thousands of blogs daily to check for posts related to searches. This action requires a longer collection cycle for all blogs and can cause a delay in the posts from this channel. That's why blogs are outside the standard collection cycle for searches. If your search is collecting posts within 1 hour, for example, it's possible that blog posts may be included later.

To ensure that a blog is part of a priority queue and enters the search with a shorter collection time, include it in the "Add blogs" field.

Why was a post in another language collected in my search?

STILINGUE's Natural Language Processing (NLP) is used to understand features of our written language and transform them into computational language. The SNLP processed for Portuguese is not the same as that used for processing English or Spanish (or other languages). Therefore, before this process occurs, it is necessary to identify the language of a publication.

When the text of the posts has more words, such as organized content from news or blogs, language identification becomes easier due to the quantity and quality of material (words) to be analyzed. For social media texts, often short and not always following spelling norms, languages can blend, making identification more complex. In addition to spelling errors, the presence of mixed publications (mixing more than one language) and cognates (expressions that exist in more than one language with different meanings) create ambiguity and also complicate identification.

For this reason, it may happen that posts in another language are included in the research. It is important to emphasize that there is no complete accuracy in this classification. Currently, STILINGUE has an accuracy of 90.95%.

Why was a post with a term from my Search Configuration not collected?

The collection of Listening includes news, reports, social media posts, and other formats that can be part of the collection in the registered survey.

For this content to be collected, it is necessary that the related websites are stored in the STILINGUE database, which already has an abundance of blogs and portals. There are over 6 million social media profiles, over 700,000 blogs, forums, and over 30,000 major online news portals.

Therefore, most of the profiles of greater relevance are already in our database, but when a profile is very specific, regional, or has a low level of interactions on the page, it does not enter this database.

By adding a profile in the Search Configuration, you contribute to this database and make more content available for collection. Terms inserted in Groups (in the case of text-based posts) are not always a guarantee that the publication mentioning the term will be collected, as some social media collections like Twitter, for example, yield sample results.

Not all data collection is done through text searches. To increase the chances of data collection, it's essential to research and add specific profiles, portals, and blogs to the search. That's why it's important to include profiles with the characteristics mentioned in your Search Configuration.

Can I create multiple (fake) pages and connect them to receive more tokens?

No. Valid proprietary pages are required to establish and maintain connections. Having a high number of pages that don't correspond to real people/institutions or lack internet activity is futile. Furthermore, this action violates social media policies.

Can I take data from a Stilingue account and use it in another account that is not connected to a page?

This procedure is not possible. You need to have the page connected to the account where the dashboard that will receive the posts is configured.

Can I use Instagram Business Discovery to monitor my own proprietary page?

No. Proprietary pages already connected to the tool cannot be registered in Listening. However, the data from proprietary pages can be accessed through the Conversation Filter within the Listening area.

What criteria are necessary for an influencer to be included in the Influencer Center?

The publishers (authors of a post) collected in the research become part of the Center when they meet one of the two criteria:

Having a follower count greater than or equal to 5000;
Having at least 5 posts in the last 30 days with an interaction count greater than or equal to 300.

What is the accuracy of the STILINGUE AI engine for sentiment classification in Portuguese posts?

First, it's important to understand accuracy as a metric that informs the overall performance of the AI. This metric is dynamic, meaning its outcome varies depending on the analyzed dataset. Currently, the accuracy for sentiment classification in PT-BR posts is between 75% and 82%. This means that 75% to 82% of the total sample was correctly classified by the AI engine.

Accuracy is determined by the analysis of different data samples. In this way, it is possible to establish a more common accuracy range in sentiment classification performed by STILINGUE's AI engine.

Some factors within the sample used, as well as the characteristics that make up Brazilian Portuguese, directly influence the percentage obtained, as the language is constantly evolving, with words being redefined and new expressions created daily.

What's the difference between placing an operator in Group or in Context?

The main difference is that the Groups will seek posts containing the terms inserted in each Group. As for the Context, it is not responsible for gathering data but rather for narrowing down the focus of the search. For example: if there are words in the search that have two distinct meanings, by placing a term in the Context, you indicate which of the meanings you want to assign to that word. Include terms that are related to your goal and can help find posts based on what has been entered in the Groups.

Another example: When you place the term "jogo" in a Group, any post mentioning "jogo" will be collected regardless of the context. There are games related to sports, cooking games, board games, and many others. It's up to the Context to specify which of these games is relevant to your research.

If I enter repeated terms with the "&&" operator, will I only collect posts that have duplicate terms?

No. Before explaining how collection works for this type of case, it's important to show the arrangement of terms using this collection operator in the Search Configuration:

The operator represented by the symbol "&&," functions like the "AND" in boolean search and only returns posts that have both or more terms entered in the post text.

For this example, in which the term "red" is also part of "red house," any post that contains the word "red," only once, will be collected. It's not necessary for the words within the post to be duplicated.

The behavior described, however, is only valid for this operator (&&). Repeated terms when the Exact operator is used, for example, will be collected in the exact way they were entered.

If I don't have a proprietary Instagram page, can I collect posts for my listening monitoring?

It's not possible to monitor Instagram hashtags without a proprietary page. Access to Instagram posts is only granted to users who have Instagram Business pages for the token (access key) authorization. This is a rule of the Instagram API, and you can learn more about this decision through the link.

For more information, visit the discussion on the subject at our community or videos on our channel. 😃

Related articles