Important: Summaries should never be the sole basis of any investment decision.
Turn this feature on under the News Feed Layout filter:
There is way too much information out there for any one person or team to understand everything. But a lot of value can be gleaned from CityFALCON's gigantic data set. Our Summarisation feature highlights the most relevant points so you can get a good overview before diving deeper into your research.
Automated Summary of FAANGM watchlist
How it works
Our machines aggregate all of the content for which you've indicated interest, then they analyse all of it based on many dimensions. The most representative content is chosen, paraphrased, and delivered to the user in an easy-to-digest summary.
Since one story can represent a cluster of tens or even hundreds of very similar stories, this representative approach greatly condenses all of the content into a few lines. As the underlying information changes, so too will the summary.
Related Content
Since we pick out representative information from a large group of similar content, we also provide the sourcing pool for that representative piece of information for users to do further research. Digging deeper into the related content can help you better understand a narrow slice of a broad summary.
Obscure topics or too restrictive filters
Some obscure topics may not produce summaries, as there may be too little content to generate one. The purpose of the summaries is to distill news into easily-digestible points, but topics with no news have nothing to distill.
A lack of content may also occur when the user sets a very narrow time period or set of publications. When the user restricts the sources and time period, our system only summarises the content that meets the filter criteria. If the filters are too narrow, there is nothing to summarise.
Information density and loss
Summarisation works because so many sources say the same thing. If thirty websites report a story, there are only so many ways to vary the headline and content while still accurately depicting the events. This usually means that a single, representative text will carry most of the important semantic value of all thirty instances.
However, when information is condensed in this way, some semantic meaning is lost. This could be the connotation of a source, which may paint the event in a slightly different light than other sources (this is where our Sentiment Analysis feature comes in handy). Another type of loss could be variation in headlines that imply conflicting information. This conflict is apparent when reading all 30 headlines, but it will be lost when condensed because it is impossible to have two ideas in the space of one.
Furthermore, since representation necessarily requires agreement, true-but-conflicting information may be obscured if it is not popular. Conversely, incorrect information may be weeded out because it does not represent the entire group, reducing the spam you must wade through.
Accuracy
Since the machines are choosing the most representative data, they may sometimes make mistakes when deciding which content is truly representative. This is a well-known issue in data science that receives a lot of research attention. The results are pretty good, but they are not perfect. And they may never be perfect. Regardless, humans make mistakes, too, and they take much, much longer to process all of the information. Moreover, because of the condensation, not all subtle viewpoints can be communicated.
CityFALCON Summaries are an entry point and provide a broad overview of the entire dataset. Because sources tend to agree and reflect reality, they are invaluable as an overview tool. But before you make an investment decision, you must dive deeper into the topic.
Missing topics
If you feel like your summaries do not include all of the topics on your watchlist, you may be correct.
First, summaries should be short. We aim to have just a few lines per summary. However, many users have watchlists consisting of 50+ topics. In those cases, due to the natural effects of information density, either summaries must expand or information must be lost - basically, you can't talk about 50 different topics in 5-10 short lines because the information simply doesn't fit in such a compact space.
If this is the case, our algorithms choose the most important information and provide that in the summary.
Second, we weight newer content more heavily in our algorithms. If you have sorted a feed by CityFALCON score, you may have older stories at the top of the feed. Because they are older, they may not be relevant to a summary right now, and thus that particular topic/story may not appear in the summary.