HonkinNews for January 17, 2017 Now Available
January 17, 2017
This week’s HonkinNews takes a look at Yahoo’s post Verizon name. No, our suggestion of yabba dabba hoo or was it “hoot” was not ignored by Yahoo’s marketing wizards. We also highlight Alphabet Google’s erasure of two letters from its “alphabet.” Goners are “S” and “T”. Palantir is hiring a people centric person. The fancy title may have an interesting spin. Two enterprise search vendors kick off 2017 with a blizzard of buzzwords. The depth of the cacaphones is remarkable because search by any other name would return results with questionable precision and recall. The featured story is the Mitre’s Corporation Jason Report. If you have an interest in artificial intelligence and warfighting, the report provides some insight into what the US Department of Defense may be considering. But the highlight of the unclassified document is a helpful description of Google’s TPU. The seven minute program is at this link. For fans of XQuery, we have a bit of input for you too. Proprietary XQuery too. The program is produced in old fashioned black and white and enhanced with theme music from the days of the Stutz Bearcat. From the hotbed of search and content processing, HonkinNews is different. We’re presenting information other big time outfits ignore. Mitre is a variant of Massachusetts Institute of Technology Research. There you go. Live from Harrod’s Creek.
Kenny Toth, January 17, 2017
OpenText Goes Cognitive
January 17, 2017
The Canadian roll up of ageing information access products has a new angle for 2017. No, it is not the OpenText “innovation” tour. No, it is not adding Alexa to its mind boggling array of content access systems such as BRS Search and IDI Information Dimension’s Basis. Not even the historical Fulcrum product gets an Alexa skill.
The future for OpenText is … a next generation cognitive platform.
Even more fascinating is that OpenText has had a Christine Maxwell moment and named the product Magellan. Who remembers the Magellan search and content access system? Ah, no one. I am not surprised. I met a millennial who had never used a camera with roll film yesterday.
If you want to know a bit more about the exciting old-new products, navigate to this OpenText page or follow this link to a video. Yes, a video because OpenText is definitely in step with the next generation of enterprise software customers.
Stephen E Arnold, January 17, 2017
Apache Tika Could Be the Google of Dark Web?
January 16, 2017
Conventional search engines can effectively index text based content. However, Apache Tika, a system developed by Defense Advanced Research Projects Agency (DARPA) can identify and analyze all kinds of content. This might enable law enforcement agencies to track all kind of illicit activities over Dark Web and possibly end them.
An article by Christian Mattmann titled Could This Tool for the Dark Web Fight Human Trafficking and Worse? that appears on Startup Smart says:
At present the most easily indexed material from the web is text. But as much as 89 to 96 percent of the content on the internet is actually something else – images, video, audio, in all thousands of different kinds of non-textual data types. Further, the vast majority of online content isn’t available in a form that’s easily indexed by electronic archiving systems like Google’s.
Apache Tika, which Mattmann helped develop bridges the gap by analyzing Metadata of the content type and then identifying content of the file using techniques like Named Entity Recognition (NER). Apache Tika was instrumental in tracking down players in Panama Scandal.
If Apache Tika is capable of what it says, many illicit activities over Dark Web like human trafficking, drug and arms peddling can be stopped in its tracks. As the author points out in the article:
Employing Tika to monitor the deep and dark web continuously could help identify human- and weapons-trafficking situations shortly after the photos are posted online. That could stop a crime from occurring and save lives.
However, the system is not sophisticated enough to handle the amount of content that is out there. Being an open source code, in near future someone may be able to make it capable of doing so. Till then, the actors of Dark Web can heave a sigh of relief.
Vishal Ingole, January 16, 2017
RAVN Flaps Amidst a Flurry of Feathers
January 12, 2017
I read “Abraaj Drives Innovation in Private Equity Market with Implementation of RAVN’s Cognitive Search Solution.” The main idea is that RAVN, a vendor of enterprise search, has snagged a customer. That’s good. What’s interesting about the write up is the language of the “news.” Here’s a rundown of the words I highlighted as flaps of the RAVN’s marketing department wings:
- Access
- Artificial intelligence and AI
- Classify
- Cognitive search
- Collaborate
- Component
- Connect enterprise
- Data mining
- Deal flow
- Differentiation
- Drive innovation
- Dynamic decisions
- Engagement
- Engine as in “cognitive engine”
- Experts and expertise
- Extract
- Functional knowledge
- Ground breaking
- Growth markets organization
- Highly distributed network
- Internal and external content
- Intelligently transforms
- Interrelationships
- Knowledge graph
- Knowledge management
- Knowledge sources
- Leverage
- Lifecycle
- Monitoring
- Multi geography
- Navigate
- Phases
- Platform
- Proprietary
- Sector knowledge.
- Sectoral
- Secure
- Solutions
- Teams
- Transformation
- Unstructured
- Visualize
What’s left out? No analytics, which is one of the must have functions for a modern search and content processing system. My hunch is that RAVN has numbers in its nest. In the millennial editing frenzy, counting visitors and other useful items was overlooked. Amazing stuff. No wonder some folks roll their eyes when enterprise search vendors trot out keyword search dressed in rhetoric honed by Sophists.
For more lingo which makes search seem more than it is, review the list of cacaphones at this link. Ah, the cacophony of search and retrieval vendors.
Stephen E Arnold, January 12, 2017
Epi-Search and dtSearch
January 11, 2017
I read about Epi-Search in “Epi-Search, A Free Academic Resource, Lets Researchers Enter Up to 10,000 Words of Text and, Using a dtSearch® Engine API, Find “More Like This” Across the ISCE Library.” The idea is a good one. Plug in a research summary or journal article abstract. The system then outputs related documents. The search plumbing is provided by dtSearch, a vendor based in Bethesda, Maryland.
I ran a test using the description of my Cyberosint: Next Generation Information Access monograph on the system at this link. The system returned 10 hits to related documents. Here’s a partial list:
Only the “Penal Populism” was in the same city as the ball park in which I kick around.
The link to Google search results was in the ball park’s parking lot. But the majority of the Google hits point to sites for cyber security, not for the use of open source intelligence to obtain operational intelligence. The Google search grabbed the notion of bad actors compromising network integrity. Important, but not the game I follow. The Google search results returned by the Epi system were PDF files and advertisements.
On a side note, there is a product called Episerer which is offered by a company called Epi. You can get information at this link. Epi’s content management system includes a search engine marketed as Find. Perhaps the name clash can be resolved?
Stephen E Arnold, January 11, 2017
Coveo Search: Reveal Engine
January 11, 2017
Enterprise search is alive and well when it comes to pivots and jargon. One example is Coveo’s adoption of the moniker “Reveal Engine.” The idea is that one can search to find needed information. If the notion is unclear, you can watch the “Coveo Reveal Engine Explainer Video.” The idea is that Coveo’s software puts intelligence everywhere. I love those categorical affirmatives too.
Coveo explains:
Meet Coveo Reveal Engine, the self-learning technology in the Coveo platform that makes intelligent search even smarter. It continuously analyzes your visitors’ click stream data and behavior patterns captured by search usage analytics, then accurately serves up the content that is most likely to drive conversions and ensure self-service success. In other words, Coveo learns which content delivers the best outcomes and then automatically tunes search results to ensure the right content always rises to the top – without any manual effort on your part. Think of it as having a built-in concierge that intuitively knows what your visitors are looking for. It makes intelligent recommendations based on a deep understanding of what has worked best for others. Coveo understands your visitors’ intent. It auto-completes search terms, provides relevant related search suggestions, and even recommends complementary content they hadn’t originally sought out.
The pivot is that Coveo is positioning its search system to handle self service questions. More information about Coveo is available at www.coveo.com.
Stephen E Arnold, January 11, 2017
Improve Your B2B Search with Klevu
January 6, 2017
Ecommerce sites rely on a strong search tool to bring potential customers to their online stores and to find specific products without a hassle. B2B based companies have the same goal, but they need an entire different approach although they still rely on search. If you run a B2B company, you might want to take a gander at Klevu and their solutions: “Search Requirements For A B2B Retailer.”
In the blog post, Klevu explains that B2B companies have multiple customer groups that allow different pricing, products, discounts, etc. The customers see prices based on allocation from the store, but they cannot use a single price for every item. Search is also affected by this outcome. Klevu came out with the Klevu Magneto plugin to:
With the help of our partners and our in-house expertise, we came up with a solution that allows such group prices to automatically work with Klevu. The Klevu Magneto plugin fetches the group prices and, at the time of showing the search results, Klevu’s JavaScript determines the relevant price for rendering. We’ve also ensure that this works in Ajax / quick search as well, as this was an important requirement.
The Klevu Magneto plugin also has an SKU search option, maintaining the same landing page within search results, and instant faceted search. Klevu researched the issues that its B2B customers had the most problems with and created solutions. They are actively pursuing ways to resolve bothersome issues that pop up and this is just the start for them.
Whitney Grace, January 6, 2017
Textkernel: Narrowing Search to an HR Utility
January 5, 2017
Remember the good old days of search? Autonomy, Convera, Endeca, Fast Search, and others from the go go 2000s identified search as a solution to enterprise information access. Well, those assertions proved to be difficult to substantiate. Marketing is one thing; finding information is another.
How does a vendor of Google style searching with some pre-sell Clearwell Systems-type business process tweaking avoid the problems which other enterprise search vendors have encountered?
The answer is, “Market search as a solution for hiring.” Just as Clearwell Systems and its imitators did in the legal sector, Textkernel, founded in 2001 and sold to CareerBuilder in 2015, , is doing résumé indexing and search focused on finding people to hire. Search becomes “recruitment technology,” which is reasonably clever buzzworking.
The company explains its indexing of CVs (curricula vitae) this way:
CV parsing, also called resume parsing or CV extraction, is the process of converting an unstructured (so-called free-form) CV/resume or social media profile into a structured format that can be integrated into any software system and made searchable. CV parsing eliminates manual data entry, allows candidates to apply via any (mobile) device and enables better search results.
The Textkernel Web site provides more details about the company’s use of tried and true enterprise search functions like metadata generation and report generation (called a “candidate profile”).
In 2015 the company had about 70 employees. Using the Overflight revenue estimation tool, Beyond Search pegs the 2015 revenue in the $5 million range.
The good news is that the company avoided the catastrophic thrashing which other European enterprise search vendors experienced. The link to the video on the Textkernel page is broken, which does not bode well for Web coding expertise. However, you can bite into some text kernels at this link.
Stephen E Arnold, January 5, 2016
Is Your Data up for Sale on Dark Web?
January 4, 2017
A new service has been launched in UK that enables users to find out if their confidential information is up for sale over the Dark Web.
As reported by Hacked in an article This Tool Lets You Scan the Dark Web for Your (Stolen) Personal Data, it says:
The service is called OwlDetect and is available for £3,5 a month. It allows users to scan the dark web in search for their own leaked information. This includes email addresses, credit card information and bank details.
The service uses a supposedly sophisticated algorithm that has alleged capabilities to penetrate up to 95% of content on the Dark Web. The inability of Open Web search engines to index and penetrate Dark Web has led to mushrooming of Dark Web search engines.
OwlDetect works very similar to early stage Google, as it becomes apparent here in the article:
This new service has a database of stolen data. This database was created over the past 10 years, presumably with the help of their software and team. A real deep web search engine does exist, however.
This means the search is not real time and is as good as searching your local hard drive. Most of the data might be outdated and companies that owned this data might have migrated to secure platforms. Moreover, the user might also have deleted the old data. Thus, the service just tells you that were you ever hacked or was your data was even stolen?
Vishal Ingole, January 4, 2017
US Patent Search Has a Ways to Go
January 3, 2017
A recent report was released by the U.S. Government Accountability Office entitled Patent Office Should Strengthen Search Capabilities and Better Monitor Examiners’ Work. Published on June 30, 2016, the report totals 91 pages in the form of a PDF. Included in the report is an examination by the U.S. Patent and Trademark Office (USPTO) of the challenges in identifying relevant information to an existing claimed invention that effect patent search. The website says the following in regards to the reason for this study,
GAO was asked to identify ways to improve patent quality through use of the best available prior art. This report (1) describes the challenges examiners face in identifying relevant prior art, (2) describes how selected foreign patent offices have addressed challenges in identifying relevant prior art, and (3) assesses the extent to which USPTO has taken steps to address challenges in identifying relevant prior art. GAO surveyed a generalizable stratified random sample of USPTO examiners with an 80 percent response rate; interviewed experts active in the field, including patent holders, attorneys, and academics; interviewed officials from USPTO and similarly sized foreign patent offices, and other knowledgeable stakeholders; and reviewed USPTO documents and relevant laws.
In short, the state of patent search is currently not very good. Timeliness and accuracy continue to be concerned when it comes to providing effective search in any capacity. Based on the study’s findings, it appears bolstering the effectiveness of these areas can be especially troublesome due to clarity of patent applications and USPTO’s policies and search tools.
Megan Feil, January 3, 2017
-
- Subscribe to Beyond Search
Feature archive
News archive
-
