Input Data Format
Senti requires that your data has "content" as the column header for the social media content you upload and want coded. You may optionally add a date in a “posted_at” column that we will use for visualizing results.
Column headers should be in Row 1, so delete any rows above the row with "content". All columns other than "content" are optional, however all columns must have a unique name. We also recommend uploading a unique identifier for every row (it can be "unique_id") in order to easily keep track of your data.
Additionally, if you want to access our timeline resource, please have a column with your date information with any of the following headers: publish_date, posted_date, posted date, published, time. The following date formats are acceptable:
Remember, dates are optional.
Senti appends the coded responses to your spreadsheet, so feel free to have as many columns in your spreadsheet as you want; Senti will simply add them to the end of the spreadsheet. There is a limit of 60KB for each row of data.
Supported file types
Senti currently accepts only .csv files. You can easily export a CSV from any spreadsheet processor such as Excel, OpenOffice or Google Documents.
To convert an excel file to .csv, in excel, select File -> Save as, select Save as Type -> CSV.
Senti allows you to design a questionnaire so that you collect relevance and sentiment in your content towards specific topics.
If you’re pulling data from twitter or any source, the first step is to be sure that "Filter for Relevance" is checked in order to make sure that every tweet is relevant to your topic. You specify your topic to filter for in the "Main Topic" box.
If all of your data is relevant to the main topic, say from a Facebook page or youtube comments, please uncheck this box. Our crowd looks at the actual piece of content to decide if its relevant or not.
We provide "Is this posting relevant to __ " template and you provide the topic itself. Please phrase your topic of interest in a general way. The relevance is judged by people without specific domain knowledge so avoid specialized jargon. Feel free to include additional explanation of the topic beyond the keyword!
The information that you enter is sent to our trained crowd.
We have our trained crowd grade every piece of content on a 5 point sentiment scale in relation to your relevance question:
- 5 = Very Positive
- 4 = Slightly Positive
- 3 = Neutral
- 2 = Slightly Negative
- 1 = Very Negative
Our crowd picks up on sarcasm, nuance, and idiomatic English (i.e. "that sh** is dope!") that automated solutions can't.
In addition to finding top-level relevance, you can ask more detailed "topic-level" questions. The questions are in the form of "Does this post talk about __" where you fill in the blank. Please also provide context around your question that a non-expert on the topic could use to answer it. Topic- level questions are limited to 300 characters.
If you wanted to know if people had seen a movie, you would write in "that the person has already seen the movie", and we display to the crowd "Does the posting mention that the person has already seen the movie?" a few other examples are:
It is important that your questions are specific as possible and accessible to general public so that our trained crowd can answer them.
An example of topic-level sentiment would be for a tweet like "I really liked the Hunger Games Book, but the movie didn't live up to my expectations." We provide a sentiment score of 2 for the movie, and a sentiment score of 5 for the book.
Gender of Author
You can also find out the gender of the content. We do this by having our crowd go to the actual posting and find information that indicates their gender, like name, profile picture, etc. In order to gather this information, your file must contain links to all the posts, and the column header for the posts needs to be “article_url”.
We send all of your content to our on-demand crowd. They answer all the questions you define using our custom web forms. We have multiple, trained contributors work on every question to ensure accuracy.
We have partnerships with over 75 labor partners that gives us access to the world's largest workforce. Find out more here.
Once your data has been completed, you can see your results either in our custom dashboard or in a .csv download.
You can click on any of the dimensions of the graph to access content that corresponds with those filters. For instance, if you click on “Very Positive”, you will see only the content that is very positive. Clicking on “Very Positive” again will show all content. Filters can be combined to see any subset of data that you want. For instance, you could click “Very Positive”, “mixing with other drugs”, and select a date range to find all the relevant content.
We provide your spreadsheet with our sentiment analysis appended to your data.
For every piece of content you upload, we have our crowd determine the relevance (as you define it), as well as the sentiment towards your specific topic on a 5 point sentiment scale for $.15.
Topic- level questions cost $.03 per for every additional topic, up to 10 topics. You can also select topic-level sentiment for an additional $.03.
For example, you upload 2000 pieces of content. You always get top-level relevance and sentiment (for $0.15). If you also select 3 topic-level questions and 2 topic-sentiment questions, your total price would be $0.30 / piece of content ($0.15 + 3 * $0.03 + 2 * $.03). So for 2,000 pieces of content, the total price would be $600.
We accept all major credit cards. All payments in Senti are processed by Stripe.
If you need to arrange for alternate payment, please contact us here.
API Coming Soon
We're planning on releasing a public-facing API in the near future. Contact us to find out more.
Senti app only supports sentiment coding in English. If you would like non-English sentiment coding, send us us an email.