Reference Reviews

Péter's Digital Reference Shelf

December 2006

Title: SiloBreaker
Publisher: Silobreaker, Ltd.
URL: http://www.silobreaker.com
Cost: $59/month or $398/year for individuals
Tested: November 22-26, 2006

The Context

There are zillions of free news sources on the Web, and millions of news aggregators. They are certainly better than the news sources on television, where in the mornings you get the chit-chats by “we-must-be-funny-at-all-cost-all-the-time” people, or the weatherman presenting and commenting on news.

The late evening news is different, but not much better, with faster and louder talking personalities yakking about the same inconsequential people and events ad nauseam, and by the top of the hour you still don’t know what the real news of the day was.

No wonder that increasingly more people are getting their news from the Web than from print newspapers and television – at least in the U.S.

For people who must be aware of and quickly understand the context and implications of the news as they start their day, the traditional news sources just don’t cut the mustard.

The news aggregators are free and many of them allow acceptable customization. There are also some good RSS services focusing on the news about certain topics or industries, and these can be subscribed to for free. But these are linear services with limited scope and very limited context even when they offer links to related stories.

Enter Silobreaker, which offers for about the price of a dollar a day, an innovative tool set that helps users be in the know and to cut through the clutter and/or the real and the pseudo news that is not pertinent to you.

The Content

Silobreaker claims to cover more than 10,000 qualified sources (this is more than twice the number of Google News), and adds 20,000 new items a day to its news silo. The producer makes it clear that the “general focus [of its services] is on global instabilities, whether political, economical or related to crime, security, the environment, health, military or industry”. Well, that’s about 80% of all the news worth knowing (if we don’t count the disproportionally huge amount of news in the entertainment/celebrity industry).

The sources are articles from journals and newspapers, items from wire services, reports from government and other agency reports from all over the world through the Moreover syndicated services. Silobreaker has a special proprietary source. It is a fact compendium, a combination of encyclopedia, almanac, gazetteer, factbook and other ready reference sources maintained by the producer, and mashed up with the news items. The majority of the news content comes from moreover.com, one of the best news aggregators, with clients like Factiva, Microsoft, British Telecom, Ernst & Young. Many other portal services also use a variety of different subsets of Moreover, but it is not directly available to individual users.

The content in and by itself is not revolutionary in 2006. What is more revolutionary than evolutionary is the way that news content is selected, processed, disaggregated, sliced and diced, analyzed, contextualized with the Elucidon software. That’s why my content description is relatively short and the emphasis is put on the software.

The Software

There are several modules in the Silobreaker software, but their beauty is that the searcher does not really feel the change and switch, because they are so well integrated that one can keep the momentum of swinging across the news canopy even when doing a small or large detour. The function of these modules can be best explained by following an actual scenario of checking the news with no specific questions in mind, just to get a feel for the events of the day.

On the day when I was creating the screenshots for this review, the homepage of Silobreaker showed this panoramic view of world events. It is no surprise that the top stories were about the poisoned Russian spy; another chapter from the endless cycle of revenge massacres between Sunni and Shiite militias in Iraq; and actions to welcome the cease-fire between by volleying some rockets into Israel from Gaza. This is also what would dominate the TV news, not as if the latter two were anything new. Scrolling down the screen shows a list of news items – not much to write home about, yet.

What is, however, also news, and you can quickly spot, is the unusually prominent appearance of the name of two relatively small and not touristy towns in Southern Thailand in the Places category right after Belfast (obviously for the assassination attempt which already was among the top places the day before). While the Thai news may not be that relevant for you and your aunt and the majority of the world population, it did get my attention as I am heading to Europe through Asia at a lower cost with a short stopover in Thailand to avoid the pains of the American and European airports.

Hovering above the town name opens a tiny pane showing the population and the geographic location of the two towns, along with small icons. Clicking on the 360 degree icon puts Pattani into focus, and rearranges the tableau, showing different entities: three Thai politicians’ name under the Persons category; then Thailand, Malaysia and Singapore under the Countries category, then three international organizations involved in crisis situations under the Organizations category, three other companies under the Companies category, and three new subject terms (Insurgency, act of terrorism and Special Forces) under the Topics category. The location of the small towns where the news-making events took place is displayed on a close-up map.

Instead of clicking on any of the entities for further information, the user needs only to hover above the entity to see a compact biography of the new prime minister, a map and population data of neighboring Malaysia, etc. These small pop-up panes have a series of icons. Clicking on them would save the biographical data about the person, filter the result set to those articles in which the name of the person selected occurs, or create a network map showing relationships between the selected entity and others. Hovering above any of the entities would often provide further details about them in pop-up panes. They have also the series of icons for the same functions described in the previous paragraph.

This mix of the entities already paints a picture of what happened but much more information can be invoked without losing sight of the context and the associated group of entities. This is just the beginning.

Headlines from newspaper and journal articles about the event appear on the left side. In the middle there is a chart about the number of articles in which the name of the entity or entities appeared.

Clicking on the connecting lines between the entities offers an option to Show Evidence for the relationship. This triggers the instant creation of a refined subset of the newspaper and journal list in which the name of all the entities claimed to be connected by the software appear. Once again, hovering above the headline will automatically display a preview pane to show excerpt form the source document.

The set can be quickly sliced by news sources, which appear on the top of the list. The 10 most productive sources for the news at hand can be displayed on a checkbox list to limit the result to the selected sources. It would be even more useful if searchers were given an option to choose their favorite sources. It may happen, of course, that none of the pre-selected sources would have a story about the event, yielding a zero item result. There is a compromise to avoid such situation by offering a list of the 10 most productive sources for the particular set of news items, and the extra item could be a check box labeled as user’s preferred sources.

There are filtering options to choose document genre, such as newspaper articles versus compact newswire items about an event, or a chronological list of events that lead to the currents situation, or statements by the person(s) involved in the event reported. This latter genre, the cluster of direct quotations, is a unique and powerful filter that I have not seen in any other system. All these filters can be gradually applied to zoom in on the special angle of the news in which the searcher is interested.

The original or the modified set can be further sliced and diced by geography, industry, people, organizations, companies and specific topics (in this case, subjects like acts of terrorism, armed conflicts, murder/manslaughter, gun crime, Buddhism, Islam). Throughout the process every items can be saved in a scrapbook, and the evolving query itself can be saved with a user-chosen label. Filters can be also set by date range and document type.

Silobreaker is not (yet) perfect. You will certainly find some oddities. For example, it does not seem to be able to handle diacritic characters well. This is odd because the partner in the project is InfoSphere, a Swedish consultancy firm in competitive intelligence, and diacritical and other special characters abound in Swedish.

Looking up the news related to Hungary, understandably brings up the name of the prime minister as the first entry under the People category – and as the third entry, not so understandably. The reason for this is that the first entry uses the diacritical mark in his last name the third entry does not. The first entry shows its mini-biography with a photo, the third one merely shows Person as an entity identifier, which should be omitted anyhow when a name appears under the People category. It is a safe bet that users of a competitive intelligence system would have basic intelligence to figure this out.

The good news is that the software seems to be handling well the other bane of searching, the various transliterations of the same entity, such as Hizballah (which is in the news every day appearing in this format in 27,429 news items), and in 970 items as Hezbollah, not mention the other variants. – But under the Organizations category the Hizballah entry seems to be the preferred term, and more importantly, the set includes not only items where the name is spelled as Hizballah, but also items where it is spelled as Hezbollah or Hezbullah. The same was true in my testing for the variant transliterations of the Lybian president, and al Qaeda.

In spite of its appeal it is yet to be seen how these differences would change the composition of the entire landscape of entities, and their connectedness for the analysis of an important issue, where the devil is in the details. This is a semantic, not merely a syntactic question or muckraking – therefore a key point of the reliability of this remarkable and innovative new system.

Opinions expressed in this review do not necessarily reflect the opinions of Thomson Gale, its employees or affiliates. We cannot guarantee the accuracy of information contained in non-Thomson Gale sites.

— Péter Jacsó