Page sections:

General questions

Who is playence?
playence is a software vendor that has engineered a Web-based solution for improving information access to multimedia content. playence offers advanced interfaces for filtering and discovering information, combined with the ability to understand natural language queries.

What does playence offer?
playence offers corporations the means to search within video and audio in much the same way we currently search within text documents. With playence, one can use natural language queries to land a search at the very second a word is being pronounced through audio or the moment a character appears on a video stream.

Who should use playence solutions?
Any business working with multimedia (i.e. video, audio, pictures and text) content and wants to improve access to those assets, either internally through proprietary applications or intranets, or externally, through dedicated Web sites.

How does playence differ from traditional keyword-based searching?
With traditional keyword-based searches, users' queries are matched one-to-one with the tags or words describing the assets. If they match, assets are considered relevant for the result set according to a ranking criteria. With playence, assets are relevant for the result set only if what you mean matches what the assets do or contain. For example, if you are looking for videos where “Michael Jordan is talking about Nike”, all assets containing the tags “Michael Jordan” or “Nike” do not necessarily represent correct results to your query. This is due to the fact that the role of those tags may or may not match the query. playence is able to understand the role of those tags and other metadata, yielding significantly better results.

What do you mean the query is expanded?
Query expansion deals with increasing the recall (i.e. number of assets in the result set) by automatically finding synonyms, similar morphological expressions, spelling errors, and of course, contextual analysis and matching.

Is query expansion the same as query refinement?
No, they are different. While query expansion affects recall, query refinement actuates on the precision. At playence we do not understand search as a one shot process.  On the contrary, we believe the user should be provided with the opportunity to dynamically polish the query and visually navigate the information structure. Query refinement is a powerful technique that allows dynamically narrowing your query based on the information you are presented with. For this, you can use the relationship viewer, probabilistic tag cloud, browsing and other filters and controls available through the application interface.

What is serendipity?
Is when one finds something that one was not necessarily looking for, but that is useful. With traditional search engines, users try word combinations to find information. You can only find it if you are looking specifically for it. With playence you can find information you do not necessarily know that exists. We will provide you hints of what we think might be relevant for you, although not your primary search goal. In this way, your search experience becomes a lot richer.

Setup and installation

Do I need to replicate my multimedia assets in a separate storage for playence to work with it?
No. playence integrates seamlessly with databases, CMS, ECM, DAMs, file storage systems, etc. There is no need to replicate content or to install additional storage. During set-up and every time an asset is added, we add relevant information to the index. Only when a user wants to reproduce or access a particular asset we retrieve it from its actual storage. Otherwise, we just work with the information on the index.

Does the playence software have to physically reside at the same place as the multimedia assets it works with?
No. Your multimedia assets can be located at one or more locations or systems and the playence software in a different one (e.g. Amazon cloud). The only restriction is that the software and the index remain together.

What installation options are available?
Depending on your needs we offer two different options. Customers can choose to install the software appliance on their data center and have their own staff administer the system, or have it installed in the Amazon cloud. With the later option there exists the possibility to have playence administer the instance for a small fee.

Does playence offer predefined thesaurus? Can I use my own one?
Yes, our software comes with built-in thesaurus for a number of domains and applications. In case you want to use your own one, that is not a problem at all, we can easily integrate it.

Security

I have security concerns and want to define different levels of access to the information on my system. Is this possible with playence?
Yes. playence allows you to define fine-grained access policies. We also offer system admins the means to define groups with specific access privileges and assign users to them.

How does playence enforces that users only access those assets for which they have security clearance.
We do that at the index level. When we index assets we also take into account users privileges. At query time, those are checked against the user access level. If they match, the user is granted access to the asset in question, otherwise denied.

My organization poses sensible data that we would rather not place in the cloud due to internal security policy. Can the playence software run in our corporate data center?
Absolutely, we deliver our solution a SaaS and also as a software appliance in your corporate data center. You chose whatever works best for your organization.

My organization prefers the SaaS, yet we do not want to be bothered with the administration of the instance running on the cloud. Is there a work around for this?
Absolutely, if you prefer the cloud option, playence can take care of administering the production environment for you.

Speech-to-Text

What languages does your speech-to-text engine support?
We support the most commonly spoken languages US and British English, Spanish in its South American and European variants, French, German, Portuguese, Chinese, Russian, Korean and many others.

What is accuracy of the Speech-to-Text engine?
It very much depends on the quality of the signal and the diction of the speaker. If those are good, we can do pretty good and easily be around 80% or more.

Can I improve its performance?
Yes, you can. Unlike other Speech-to-Text technologies that require the development of complex language modules to improve performance, with our technology it is as easy as feeding the system a document with new words or expressions. It will learn from them and recognize them in the future.

How long does it take to generate the transcription of an audio track?
It depends on the level of service required and configuration chosen. It ranges from real time to 1/8th of real time.

Can the transcription run as a batch process?
Yes, it can.

Can your technology identify who is talking?
Yes. You can configure the voice registries of the persons you would like the system to recognize and we will tell you who is talking at each moment.

Mind Maps or Information Maps

Are information maps or mind maps the same as ontologies?
Information maps are term a lot more familiar to most of the people than ontologies. Although not scientifically exact, we use them interchangeably, as they both can be used to model relations among concepts and information’s structure.

What is the difference between and ontology and a thesaurus?
An ontology is a vocabulary used to represent knowledge in a given domain using a set of concepts and the relationships among those concepts. A thesaurus lists words together according to similarity of meaning (i.e. synonyms, antonyms and hyperonims).  We support both.

What is the difference between an ontology and a taxonomy?
While taxonomies only support parent-child among concepts, also known as “is-a” relationship, ontologies allow linking concepts using any relevant verb. As a consequence ontologies offer much richer expressivity and are a lot more powerful than taxonomies.

What is the difference between and ontology and a knowledge base?
In short, ontologies provide the vocabulary that can be used to define a knowledge base. The same languages used to define ontologies (i.e. OWL or RDF) can be used to define knowledge bases. An ontology provides information about what is relevant in a domain. A knowledge base provides information about individuals in that domain.

Does playence support multiple information maps for the same or different users?
Yes, it does. Each user can have one or more information maps associated. In the user settings section you are able to specify which one to use for each search process depending on your interests and current view of the world.

What is the benefit of supporting several information maps?
They allow users with different backgrounds and interests access information using their own different vocabularies and views of the world. For example, a video produced by Customer Care regarding common user requests contains valuable information for Sales and Engineering. However, each one is interested in different views of the same information. By using different information maps, Sales and Engineering can find the relevant information using their own vocabulary.

What happens if I want to evolve my information map, can I easily do it?
Yes, it is vey simple. You can do it yourself or let your admins do it for you. It all depends on your privileges level and your confidence with the information structure.

Upload & Indexing

Does playence support bulk upload and indexing?
Yes it does. Also as a batch process.

Does playence support single asset upload and indexing?
Yes, it does, if your privileges have been set to do so.

How long does it take to index an asset?
It depends on the size of the asset, typically tenths of a second.

Does it take the same amount of time to index small assets as large ones?
No, it doesn't. The time needed for indexing a document of 100 pages is bigger than the time needed for indexing 10 documents of 10 pages. Performance, in this case, is not linear with respect to size.

When indexing, does the system take into account tags, closed caption, transcription and free text?
Yes, we index all that metadata.

What is the role of the information maps during the indexing process?
They provide the structure of the information used to find relevant relationships in the assets and to classify them in the right part(s) of the information map for later retrieval.

Can playence automatically generate tags?
Yes, we can. We are able to understand what is happening on the asset and based on its contents suggest the most relevant set of tags that accurately describe it.

Manual annotation

Does the system support tags, transcriptions and free text?
Yes, it does. The solution can be easily customized to support different types of metadata.

Does playence complement the automatic indexing process with manual annotation?
Yes, it does. Once assets have been uploaded and automatically indexed, users depending on their privileges, can add or modify existing annotations. Also enter free text or tags as desired. When done, modifications are stored and indexed, as needed.

What is dynamic rating?
The traditional 5 stars rating provides a general overview of the quality of a multimedia asset. However, portions of a video or audio might be more relevant than others. Dynamic rating is a novel technology we have incorporated to our solution that allows defining which portions of a video or an audio are the most relevant based on users’ input. Through a control that slides up and down users are able to rate multimedia. The result is a graph that provides a detailed overview of what users think are the most interesting portions of a video or an audio.

Search & Browsing

What is the role of the information maps during the browsing process?
They provide the structure of the information that users navigate.

The tag cloud and the relationship viewer change when I change the number of results per page. Is this normal?
Yes, it is. They are both calculated based on the results displayed on each page to improve their efficiency. Every time a new filter is applied or the query changes they are recalculated to provide you with the most up-to-date and relevant information. This way the precision of the search process is greatly improved.

What are cross-language queries?
Cross-language queries are queries formulated in a language different from the one used to annotate a multimedia asset. For example, we are able to correctly retrieve assets in English language through a query in Spanish and vice versa.

What are conceptual queries?
Queries formulated on natural language that using the information map as a basis, are able to understand the user request and correctly retrieve assets from the repository. For example “Pau Gasol playing with L.A. Lakers but not against Boston Celtics” is a conceptual query. The software processes and understands the query and retrieves the assets that match it.

Licensing

What type of licensing do you offer?
We offer multiyear usage-based licensing. Fees are calculated based on the number of assets under management, gigabytes in the repository or number of users, whatever suits your installation best. Multiyear licenses include Maintenance and Support. You get all Updates & Upgrades throughout the life of the license for free.

Does playence offer permanent licenses?
Yes we do. With permanent license Maintenance and Support is charged separately. Updates & Upgrades are still provided free of charge while maintenance and support fees are covered.  

What about instance administration?
For a small fee playence will take care of administering your instance if you opt for the SaaS option.