Tracking investor activities with mergeflow.stream

Posted on by

mergeflow.stream now features improved recognition of financial institutions.  Investors, venture capitalists, private equity investors and other financial institutions are now available as a new object class (e.g. in the “objects & events” tab).

This new feature allows tracking investor activities in certain technology fields, for instance.  Consider the following example screenshot (yellow nodes: technology fields, pink nodes: financial institutions):

This graph was extracted based on news from a two-week time window.  The graph shows relations between several investors and technology fields.  For instance, there is a group of investors related to “big data” and “unstructured data” (FirstMark Capital, Greycroft Partners, and IA Ventures, at around 6 o’clock in the graph).  This group was extracted from news articles reporting on a $15mio financing round at NewsCred (http://www.newscred.com/).  One of these news articles is at PEHub (http://www.pehub.com/191725/191725/).

 

 

 

How to create web feeds for public Evernote notebooks

Posted on by

Evernote (http://evernote.com/) offers web feeds for public notebooks.  Here is a description of how to create them:

https://support.evernote.com/ics/support/KBAnswer.asp?questionID=2181&hitOffset=135+29+11+2&docID=23937

You can, for instance, set up a public notebook, create a feed for it, and add this feed to one of your mergeflow topic profiles.  In these topic profiles, you can then use your notebook entries as “topic anchors”, for instance.  In order to do this, go to a document from your Evernote notebook feed and click on “document context”.  This shows you if there is topically related material anywhere in your topic profile.

Patent analyses with mergeflow.stream

Posted on by

mergeflow.stream now features improved patent analytics.  These analytics extract inventor-company-networks from patent documents and then visualize them in interactive graphs.

Inventor-company networks can be relevant for several reasons, for example:

  • Co-inventor networks often reflect how teams in a company are organized.  If several people are co-inventors, it is likely that they work together as a team.  The patents of these co-inventors usually are topically related as well.
  • Connections between companies may reflect cross-company collaboration.  E.g. if two companies are co-assignees of a patent, these two companies probably are involved in some type of collaboration (research and development partnership, joint venture, etc.).
  • Some inventors are more “central” in a co-inventor network than other inventors (“central” = have stronger connections to other inventors, or be connected to more inventors than others).  These central inventors usually play an important role in a company’s research and development efforts.
  • mergeflow.stream can analyze patents and non-patent documents in the same context.  For instance, a user can create one topic profile with patent update feeds, feeds from scientific and technological publications, and feeds with investor activity news.  This topic profile then provides a unified view on all these different types of sources.
Below are some examples for patent inventor-company networks.  To keep things simple, we limit ourselves here to patent documents and do not include other, non-patent, information streams.  All data are from patent update feeds from worldwide patent offices.
The screenshot below is for smart grid and smart meter patents.  It shows the inventors at one organization, “Electronics and Telecommunications Research Institute” in Korea (http://www.etri.re.kr/eng/).
(click on the image for a larger view)
The graph shows co-inventor clusters, the strongest cluster at the top (around Soo Kwang Kim).  Quite possibly the members of this cluster form a team at their research institute.
The screenshot below is based on patents related to speech processing technology.  It shows a co-inventor network at Philips.
(click on the image for a larger view)

Similar to the smart meter and smart grid example, this graph shows distinguishable co-inventor networks (three in this case).  Furthermore, in the graph, stronger lines reflect more common patents.  For instance, in the co-inventor network in the lower right, two inventors are more closely related than others: Eric Cohen-Solal and Michael Chun-chieh Lee.  In their network, these two inventors are more central than the other inventors.

Web feeds from Google+

Posted on by

Feed+ is an app for Google Chrome that generates feeds for Google+ (for publicly available content).  You can get Feed+ here:

https://chrome.google.com/webstore/detail/feed%20/eaiadkmgpppkonfodnepmoknpaphnlij?hl=en

With Feed+, you can generate search feeds (i.e. get Google+ updates for a certain search query), or get publicly available updates from Google+ user accounts.  You can then subscribe to these web feeds with mergeflow.stream.

 

Checklist: search feed URL generator

Posted on by

The “search feed URL generator” checklist helps with using the search feed URL generator.  The search feed URL generator can be found in the “manage” view of a topic profile.  A version for users who are not logged in is available here: http://mergeflow.net/searchsite

Get the latest version of the checklist as PDF: mergeflow.stream_SearchFeedURLGenerator

mergeflow.stream — easier search feed generation

Posted on by

We added a new feature to mergeflow.stream, a search feed generator.  There are two versions of the search feed generator:

Search feed generator for users who are not logged in to mergeflow.stream

Users who are not logged in or have no mergeflow.stream account can use this site:

http://mergeflow.net/searchsite

 

Search feed generator for users who are logged in to mergeflow.stream

Users who are logged in can switch to the “manage” view of a topic profile and generate search feeds there.  This provides more options:

  • there are previews of current feed results
  • from feed result previews, you can add the feed to the topic profile

Cyber security — following news on Operation Red October

Posted on by

On Jan 14 2013, Kaspersky Labs published a research report that analyzes Operation Red October.  Part 1 of the research report can be found here:

http://www.securelist.com/en/analysis/204792262/Red_October_Diplomatic_Cyber_Attacks_Investigation

According to Kaspersky, Operation Red October appears to be a relatively large-scale cyber-espionage operation.

We have a topic profile on mergeflow.stream that takes as input hundreds of cybersecurity-related information streams.  You can find this topic profile here:

http://mergeflow.net/profiles/research/cyber-security/docs

In order to filter for news pertaining to Operation Red October, go here:

http://mergeflow.net/profiles/research/cyber-security/docs?tagtype=rising&q=Red%20October

Starting from there, you can try out different view settings, e.g. to see tag clouds consisting of relevant objects (e.g. companies) or countries that mergeflow.stream identified in the news on Operation Red October.  You can also explore an interactive graph of objects and countries related to Operation Red October:

http://mergeflow.net/profiles/research/cyber-security/graph?tagcount=500&q=Red%20October&tagcat=Company|Country|GovernmentAgency|Organization|Software|Technology

If you want to dig deeper, you can use mergeflow.stream’s URL syntax.  A documentation of the URL syntax is here:

http://www.mergeflow.com/Content/mergeflow.stream_URLSyntax.pdf

Technology scouting: case study on GPGPU

Posted on by

– get the full case study here as a PDF –

In order to illustrate how mergeflow analytics software can be used for technology scouting, we have written up a case study.  The topic of the case study is GPGPU (General Purpose Computing on Graphics Processing Units).  GPGPU refers to the utilization of GPUs for computing tasks that traditionally are done by the CPU (Central Processing Unit).  More information on GPGPU can be found at Wikipedia, for instance: en.wikipedia.org/wiki/GPGPU

Applications of GPGPU include data processing for medical applications, genomics, proteomics, simulations, scientific computing, graph analysis, machine learning, cryptography, finance, and “big data” in general.

For our case study, we analyzed several thousand text documents (patents, scientific publications, research project descriptions, etc.) that relate to GPGPU.  We used our analytics software, mergeflow.stream and mergeflow.engine.  Here are some of our key findings:

  • From research project descriptions, we could identify several companies working on multiple-application GPGPU technologies.  Such companies may be particularly interesting to investors (e.g. venture capitalists).
  • We automatically extracted a network of GPGPU applications that shows how these applications relate to each other.
  • Among others, we found applications of GPGPU in genomics (motif detection), cyber security (intrusion detection), energy (power plant data analysis), and pricing and finance (options pricing).
  • In the data set we analyzed, machine learning is one of the R&D activities with greatest actionable relevance to many different areas of application.

Our case study shows that by using mergeflow.stream and mergeflow.engine, even non-domain-experts can quickly gain relatively in-depth knowledge of a technology field. We believe that this is mostly due to the following capabilities of mergeflow analytics software:

  • Easily collect, monitor, and analyze information from hundreds or thousands of completely customizable sources. Users do not have to rely on a fixed set of mainstream sources or search hundreds of specialized sources individually. This is essential to tasks such as technology scouting that require analyzing and monitoring information from disparate, non-mainstream, and highly specialized sources.
  • Automatically extract topic fingerprints for documents (single documents or document collections). Topic fingerprints are objects and keywords and the relations between them that constitute a topic. Topic fingerprints help users quickly discover the structure of topics, as well as how different topics relate to each other. This applies not only to topics pre-defined by the user but also to topics that are discovered automatically by mergeflow.stream or mergeflow.engine.