How to Choose a Provider to Outsource Text Annotation

How to Choose a Provider to Outsource Text Annotation

Annotation, as we perhaps know, is the act of adding information to data that makes it meaningful and easy to understand. In other words, we enhance the data by annotating it. When this enhancement is done for data that is textual, we refer to it as ‘text annotation.’

But this is in English.

The reference point here is enhancement of data so that it can be understood by computers. It is generally done as an input to building an Artificial Intelligence (AI) engine through a process of Machine learning (ML). What do we mean by ‘text annotation’ in this context?

Let us take an example. Of Resume parsing. The context here is employment.

Traditionally, the first step, when an employer sought to fill a position, was to request interested candidates to submit a Resume. The Resume, once submitted, would be perused by the HR team to ascertain suitability based on which it would be taken forward for further processing. This has been a manual, resource-hungry task for decades. With increasing volumes and pressure on performance, employers looked for technology solutions.

As technology developed, in order to make processing more efficient, employers developed a database into which interested candidates could directly input their particulars, that would otherwise be submitted by means of a Resume. While this created efficiency for the employer, as they were able to sort, evaluate and process much faster and more accurately, it created additional work for the candidate. She had to input the information already present on her Resume, into a new database. And perhaps as many times as the jobs she was applying to.

Enter AI. Leading to the evolution of Resume Parsing.

What is the process now?

Instead of keying in information field by field on the employer database, the candidate has to upload a copy of her Resume. The employer has a Resume parser at work which reads the information on the Resume and fills up the database fields by itself, only for a final assent by the candidate that she agrees with the final outcome. It is perhaps limited to large employers currently but adoption is rapidly increasing.

This has been made possible through ML for which inputs have been created with the help of text annotation services. Human text annotators have annotated thousands and millions of Resumes to identify what part of the text should be placed in which field of a manpower database. After being trained with enough data and with enough variations, the AI engine used by the employer can read the next Resume that gets submitted, and correctly classify it, even though it has not been annotated by a human being.

This is the contribution of text annotation services to AI and ML. In more technical terms, it can be referred to as the addition of metadata to text so that it becomes meaningful for machines to read (or view) and understand. And, of course, it needs to be done by human hand.

Text annotation being a critical input for AI and ML, for an outsourcer, it is important to ensure that they have the right and reliable partner to outsource text annotation.

The ensuing paragraphs suggest a set of criteria based on which the selection should be made and questions that should be asked during the evaluation process. This would enable you to separate the grain from the chaff and select suitable provider/s out of a long list.

To make the process simple, for each criteria, a set of questions is listed that an outsourcer should be asking vendors interested in providing text annotation solutions to them. These are neither mandatory nor sequential but meant as an indicative set to display the range that you might need to cover during the evaluation process. Depending upon the work that is being outsourced, you may need to shift your emphasis, leave out questions that may not be relevant and, at times, perhaps even go beyond this set.

So, here we go.

 

Domain and Functional Knowledge

  • Do they possess prior experience of doing similar work for other clients?
  • Have any of our competitors, or any other company from our industry, outsourced text annotation to them in the past or are currently sourcing?
  • Will an existing client be willing to certify their ability?
  • Will they be in a position to do our work accurately?

oWorkers has successfully provided a wide variety of text annotation solutions for global clients over the last eight years, covering a wide range of annotation types and services. It is no surprise that we count several unicorn marketplaces amongst our valued clients.

 

Depth of Coverage and Experience

  • Text annotation solutions have applications in a wide range of industries like Medical science, Aeronautical, Robotics, Agriculture, Retail, Self-driving cars, etc. To what industries have they been exposed?
  • Which of the following text annotation techniques are covered by their capability?
    • Text Categorization
    • Semantic Annotation
    • Phrase Chunking
    • Entity Linking
  •  Does their capability extend to annotations of various types like grammatical, sentiment, mood, etc.?

With oWorkers, you get proficiency in a wide variety of annotations, like text categorization, linked entities recognition, grammatical & discourse analysis and review & sentiment analysis. Being a pure play data entry BPO allows us to focus on this domain and offer industry-leading quality and accuracy. Our multiple industry coverage is transparent and proudly displayed on our website. 

 

Quality and Accuracy

  • If we outsource text annotation to them, will they have a process of sample monitoring done by a team of Quality Analysts (QAs) based on which coaching is provided to agents?
  • Will this be an independent team or within the delivery structure?
  • Do they have a Quality Control (QC) process in their processes?

With a mix of QA (Quality Assurance) and QC (Quality Control) processes and a team that is independent of delivery, so that results are not compromised, oWorkers delivers best in class performance at over 98% accuracy, supported by technology tools through our partnerships with leading providers.

 

Speed and Turnaround

  • What turnaround time are they willing to commit for delivering text annotation solutions?
  • Do they have a 24×7 delivery model? Will it be an extra cost if we were to opt for a 24×7 delivery model?
  • Which global time zones do they operate from? Can it enable us to benefit from the difference between our time zones?
  • Will they be willing to transfer existing skills with this kind of work to our project which can help us get off the ground faster?

oWorkers prefers to deliver fast, so that we can do more for you. We have three global delivery centers that can deliver to exacting turnaround time expectations, including overnight processing. Our business runs 24×7, 365 days of the year. You decide what speed you want.

 

Pricing

  • What is the price point for providing text annotation services?
  • Is the price per unit of time or per unit of output? Can they give us a choice between the two?
  • What is included in the price?
  • What is not included in the price? In other words, what is it that we will need to pay for over and above this price?
  • How much will it enable us to save in comparison to our existing cost?
  • How does their price compare with other vendors bidding for our project?
  • Are they ‘off the ballpark?’ In other words, is their quote vastly different from those of others? If yes, what could be the reason?
  • Is the price too low for them to be able to make it profitable? What guarantees should we take so that they do not ignore this work? What is the justification for the low price?

With a saving potential of upto 80% of current cost, especially if you are in the US or Western Europe, not working with oWorkers may require more justification than working with. Savings can translate to about USD 3 per hour for projects with scale. oWorkers also gives you a choice between per unit of output and per unit of time pricing.

 

MultiLingual Capability

  • How many languages are they in a position to provide support for in text annotation services?
  • Which are these languages?
  • How long does it take to add resources for non-core languages?
  • Will they be able to add a language currently not covered? How long will it take?
  • What is the depth of these skills? In other words, how many people for each language can they provide?

With support for over 22 languages across our three global centers for a wide variety of data services, the full power of oWorkers language capability is available when you outsource text annotation projects to us.

 

Technology and Data Security

  • If we outsource text annotation to them, will they be able to work on our technology or can they only work on their own tools?
  • What technology tools do they intend to use for our work?
  • Have they connected with client systems in the past or only worked on their own technology?
  • What file formats, data exchange protocols will be required?
  • Do they have an ISO (27001 and 9001) certification?
  • Are they GDPR compliant?
  • How secure are their work from home arrangements for staff?
  • What security measures do they implement to isolate the network and servers for our processing?
  • Can they implement physical access control?

oWorkers leverages the best technology solutions through its partnership with leading text annotation tools (pure data labelling tools or text annotation platforms) both on NLP and Computer Vision projects. As mandated by our ISO (27001 :2013 & 9001:2015) certification, our employees sign and operate under an NDA and work in monitored facilities with strict security protocols. It helps to have 85% of our clients as tech companies.

 

Scalability and access to human resources

  • Do they use employees or freelancers and contractors for doing work?
  • What is the depth of resources in their catchment area? In other words, what is the annual availability of fresh resources of the profile who work for them?
  • Do they have direct access to educational institutions from where they do campus hiring?
  • What do their employees rate them on Glassdoor?
  • What does the screening process for hiring look like?
  • What is the scaling up volume they can handle per day?
  • Do they offer flexibility to staff to work from home if and when required?
  • How do they equip and train the new hires for work on a project?

oWorkers relies on employees for delivery on client projects, not freelancers or contractors. We are deeply rooted in the communities we work with and draw resources from. With a 4.6/ 5 rating on Glassdoor, we remain a preferred employer in each of the three locations we operate from. Our deep connect enables us to ramp up and down rapidly, typically by about a hundred in 48 hours, for computer vision projects. We have a stringent hiring process that includes, in addition to educational background, EQ, IQ and Personality tests.

 

Financial health

  • To take on as well as outsource text annotation are decisions of responsibility and require commitment and financial strength to execute, especially for the provider. For starters, are they a profitable enterprise?
  • How many years have they been profitable?
  • Are their regulatory requirements in place and updated?
  • Where do they pay taxes?
  • Are warning signs visible on their Balance Sheet that could lead to possible curtailment of operations in future?

oWorkers is registered locally in all the global centers it operates from and has been a consistently profitable enterprise. It also pays local and social taxes for its employees and is deeply rooted in the communities it operates in.

 

General Management

  • When you outsource text annotation, it is important to assess the provider management’s commitment to the project. How active is the senior management in the day-to-day operations?
  • Does the management team have hands-on experience of the kind of work we propose to outsource?
  • Do they have centers in multiple geographies for delivery?
  • Are they in a position to offer a Business Continuity Plan to support our work by switching it to an alternate location in case the primary location cannot be accessed?
  • Do they possess a ‘project management’ framework and commitment?
  • Do they have processes through which support services are delivered so that delivery can continue unaffected?
  • Are they GDPR compliant?

oWorkers has a management team with over 20 years of hands-on experience in the outsourcing industry. They are hands-on and involved in client discussions. We have the capability to provide multi-site delivery for your projects with a common line of communication. When you outsource text annotation work to us, your business enables us to create positive social and economic change by creating employment opportunities in the marginalised communities where we work.

Recommended Posts