It is important to understand the common challenges in seeding so that appropriate
approach can be planned and necessary guidelines and processes be put in place. It is
important to note that the challenges being described in this section cover the overall picture
comprising of various service delivery databases across many states as well as multiple
seeding initiatives using varied channels being undertaken and so some of those are generic
while others are specific to a particular channel. Below are the some of the challenges that
seeding initiatives have been faced with:
1. Quality& Availability of Beneficiary data in Service Delivery databases:
Beneficiary databases have multiple data quality issues. Some of the most common
data quality issues are: errors in the data such as misspelt names/ wrong date of
births, out of date addresses etc. Beneficiary databases are yet to be digitized in
some cases, are distributed across multiple sources (such as district-wise etc.) with
duplication within and across sources, missing fields of KYR or photo data which is
very often because the particular service scheme does not mandate those fields etc.
2. Quality of UIDAI Enrolment KYR Data: Considerable improvements in the quality of
the master pincode database (address master which the enrolment client uses),
multiple versions of the enrolment client with improved features to ensure data quality
and transliteration have evolved since enrolments began. This has resulted in
significantly improved quality of enrolment KYR data in later phases of enrolments. In
earlier enrolments however there are cases of misspelt district names, wrongly
mapped pincode to VTC, redundant address fields, poor transliteration etc. Many
seeding initiatives currently are focused on geographic areas where high percentage
of enrolments have been done which naturally uses early enrolment data (early
enrolments = higher percentage of enrolments).
3. Availability of UIDAI Enrolment KYR Data: Availability of the UIDAI KYR resident
database (SRDH or equivalent) is critical to seeding initiatives especially for the
inorganic seeding channels and is also critical for verification of seeding irrespective
of channel. Currently very few States have a production quality (accessibility &
performance) resident KYR database with most available enrolment KYR data
available.
4. Quality& Availability of UIDAI Enrolment KYR+ Data: KYR+ data is not collected
in all States and among States that do collect KYR+ fields, different States collect
different sets of fields. In all cases, KYR+ data fields are optional during enrolments.
Even in States where KYR+ data is collected, very often residents have opted not to
provide the information and when provided the data is often found unreliable
(erroneous).
5. Language of Beneficiary data in Service Delivery databases: Many service
delivery databases in States are in the local language. However the software tools
that compare the SRDH KYR data to beneficiary KYR data are usable only for
English data.
6. Mobilization of Residents: Many organic seeding channels require the mobilization
of residents. Mobilization of residents is difficult, tedious and causes inconvenience
to residents.
7. Software & Hardware Infrastructure: Large scale seeding initiatives across multiple
service delivery databases is the need of the hour to expedite Aadhaar enabled
service delivery. This requires fairly significant hardware and scalable high
performing software. Most importantly it requires easy to use tools and availability of
UIDAI – RASF – Introduction Document
Copyrights © 2012. All rights reserved. Page 12 of 19
the data through the software for both seeding and seeding verification efforts in the
field to multiple stakeholders.
8. Change Management: One of the most significant hurdles in seeding initiatives
currently is many misunderstandings around the concept of seeding and lack of
clarity in the way forward. Some of the common clichés include:
• “Tool ‘X’ does 65% seeding, how much does tool ‘Y’ do?”
– Comparisons between tools is valid only if both tool ‘X’ and ‘Y’ is being used
on
- Same set of UIDAI KYR data (because % seedable is a factor of
availability)
- Same department data (because % seedable is a factor of
quality/quantity of department data which varies state to state and
among departments in state)
• “Accuracy cannot be guaranteed by seeding algorithm ‘X’”
– Accuracy cannot be guaranteed by any seeding algorithm. Our recommended
strategy is to bring Aadhaar numbers into department data marking them as
unverified and putting in place appropriate process to verify through operators
or at transaction point with beneficiary.
– This additionally allows resident to check Aadhaar KYR data and update at
CIDR if necessary (using update channels). If resident verifies seeding and
confirms Aadhaar data correctness, department can over-ride existing
department KYR data (cleaning)
• “Residence presence is required for seeding verification, so algorithmic seeding is
useless”
– Mobilizing residents for seeding is not an easy exercise (note KYR+ collection
success/ failure where mobilization was due to enrolment)
– Is 100% accuracy of seeding a must? Answer should depend on particular
transaction with department for a given scheme. 80% seeding with 1% errors
vs. 20% seeding with 0.1% errors argument.
– Resident presence just for seeding might not be resident-centric, should at
least be for all departments in one visit
– Resident presence although helpful is not necessary. Resident self-service
channels such as SMS and Online portal can be leveraged.
• “So how many duplicates/ bogus records have seeding removed?”
– It should be noted that only when Aadhaar is mandated for a scheme within a
geographical area can there be removal of duplicates/ bogus etc. If a
database has beneficiary records without Aadhaar numbers, there could be
duplicates among themselves or a duplicate of a record with Aadhaar
number. They could also be bogus records. Seeding searches can indicate
duplicates but department will have to reach out and check before removing
• “We should be using bio-auth to ensure seeding accuracy”
– Bio-auth can only tie the Aadhaar number to the person. Seeding is trying to
tie the Aadhaar record to the department record. You could very well tie an
authenticated Aadhaar record to the wrong department record (where KYR
data looks similar).
approach can be planned and necessary guidelines and processes be put in place. It is
important to note that the challenges being described in this section cover the overall picture
comprising of various service delivery databases across many states as well as multiple
seeding initiatives using varied channels being undertaken and so some of those are generic
while others are specific to a particular channel. Below are the some of the challenges that
seeding initiatives have been faced with:
1. Quality& Availability of Beneficiary data in Service Delivery databases:
Beneficiary databases have multiple data quality issues. Some of the most common
data quality issues are: errors in the data such as misspelt names/ wrong date of
births, out of date addresses etc. Beneficiary databases are yet to be digitized in
some cases, are distributed across multiple sources (such as district-wise etc.) with
duplication within and across sources, missing fields of KYR or photo data which is
very often because the particular service scheme does not mandate those fields etc.
2. Quality of UIDAI Enrolment KYR Data: Considerable improvements in the quality of
the master pincode database (address master which the enrolment client uses),
multiple versions of the enrolment client with improved features to ensure data quality
and transliteration have evolved since enrolments began. This has resulted in
significantly improved quality of enrolment KYR data in later phases of enrolments. In
earlier enrolments however there are cases of misspelt district names, wrongly
mapped pincode to VTC, redundant address fields, poor transliteration etc. Many
seeding initiatives currently are focused on geographic areas where high percentage
of enrolments have been done which naturally uses early enrolment data (early
enrolments = higher percentage of enrolments).
3. Availability of UIDAI Enrolment KYR Data: Availability of the UIDAI KYR resident
database (SRDH or equivalent) is critical to seeding initiatives especially for the
inorganic seeding channels and is also critical for verification of seeding irrespective
of channel. Currently very few States have a production quality (accessibility &
performance) resident KYR database with most available enrolment KYR data
available.
4. Quality& Availability of UIDAI Enrolment KYR+ Data: KYR+ data is not collected
in all States and among States that do collect KYR+ fields, different States collect
different sets of fields. In all cases, KYR+ data fields are optional during enrolments.
Even in States where KYR+ data is collected, very often residents have opted not to
provide the information and when provided the data is often found unreliable
(erroneous).
5. Language of Beneficiary data in Service Delivery databases: Many service
delivery databases in States are in the local language. However the software tools
that compare the SRDH KYR data to beneficiary KYR data are usable only for
English data.
6. Mobilization of Residents: Many organic seeding channels require the mobilization
of residents. Mobilization of residents is difficult, tedious and causes inconvenience
to residents.
7. Software & Hardware Infrastructure: Large scale seeding initiatives across multiple
service delivery databases is the need of the hour to expedite Aadhaar enabled
service delivery. This requires fairly significant hardware and scalable high
performing software. Most importantly it requires easy to use tools and availability of
UIDAI – RASF – Introduction Document
Copyrights © 2012. All rights reserved. Page 12 of 19
the data through the software for both seeding and seeding verification efforts in the
field to multiple stakeholders.
8. Change Management: One of the most significant hurdles in seeding initiatives
currently is many misunderstandings around the concept of seeding and lack of
clarity in the way forward. Some of the common clichés include:
• “Tool ‘X’ does 65% seeding, how much does tool ‘Y’ do?”
– Comparisons between tools is valid only if both tool ‘X’ and ‘Y’ is being used
on
- Same set of UIDAI KYR data (because % seedable is a factor of
availability)
- Same department data (because % seedable is a factor of
quality/quantity of department data which varies state to state and
among departments in state)
• “Accuracy cannot be guaranteed by seeding algorithm ‘X’”
– Accuracy cannot be guaranteed by any seeding algorithm. Our recommended
strategy is to bring Aadhaar numbers into department data marking them as
unverified and putting in place appropriate process to verify through operators
or at transaction point with beneficiary.
– This additionally allows resident to check Aadhaar KYR data and update at
CIDR if necessary (using update channels). If resident verifies seeding and
confirms Aadhaar data correctness, department can over-ride existing
department KYR data (cleaning)
• “Residence presence is required for seeding verification, so algorithmic seeding is
useless”
– Mobilizing residents for seeding is not an easy exercise (note KYR+ collection
success/ failure where mobilization was due to enrolment)
– Is 100% accuracy of seeding a must? Answer should depend on particular
transaction with department for a given scheme. 80% seeding with 1% errors
vs. 20% seeding with 0.1% errors argument.
– Resident presence just for seeding might not be resident-centric, should at
least be for all departments in one visit
– Resident presence although helpful is not necessary. Resident self-service
channels such as SMS and Online portal can be leveraged.
• “So how many duplicates/ bogus records have seeding removed?”
– It should be noted that only when Aadhaar is mandated for a scheme within a
geographical area can there be removal of duplicates/ bogus etc. If a
database has beneficiary records without Aadhaar numbers, there could be
duplicates among themselves or a duplicate of a record with Aadhaar
number. They could also be bogus records. Seeding searches can indicate
duplicates but department will have to reach out and check before removing
• “We should be using bio-auth to ensure seeding accuracy”
– Bio-auth can only tie the Aadhaar number to the person. Seeding is trying to
tie the Aadhaar record to the department record. You could very well tie an
authenticated Aadhaar record to the wrong department record (where KYR
data looks similar).
No comments:
Post a Comment