- Taiwan’s National Library of Public Information has deployed customized technology from EPC Solutions that tracks when copies of each of its 600 journals are removed from the shelf and read.
- The UHF RFID and AI based solution will enable the library to make strategic decisions related to which subscriptions to keep, increase or eliminate, based on reading habits of their patrons.
If libraries want to know how the periodicals on their shelves are being received, they usually have to rely on personal observation. Some magazines may be read again and again by patrons, others stay untouched on the shelves.
Librarians often only know how popular a magazine is based on how often they have to re-shelve it, or whether pages are getting dog-eared from handling. And because magazines aren’t checked out, there’s no digital record, as there is with books or other media.
Taiwan’s National Library of Public Information wanted to bring visibility into its hundreds of magazines on display, as it digitizes its periodical system. Its goal was to understand what magazines were most popular, which were not being read, and therefore how they could adjust subscriptions to save costs, or better serve their patrons.
Gaining Intelligence as Periodicals Go Digital
This summer they enlisted the support of RFID technology company EPC Solutions to digitally track these physical magazines and their readership.
“The RFID monitoring system can provide us with data so that we can make better decisions to subscribe to magazines that people like to read,” said Kevin Fu, RFID project leader for the National Library of Public Information. “It also allows us to better control our budget.”
The library is located in South District, Taichung, and has been in operation since 1923. In June of this year, it launched its Digital Industry Administration task force to guide a transition into digital technology. As part of the process the library system has been focused on understanding how the hard-copy periodicals it purchases are used by patrons, how often.
The five-story library has a periodicals section with 53 individual display racks containing a total of 600 different journals. The library has 47,183 volumes that must be tracked including those that are no longer on display.
They came to EPC Solutions to deploy an automated solution that would help identify when each magazine was handled and for how long.
Understanding Reader Behavior
EPC Solutions customized a UHF RFID solution with antennas at the magazine racks and tags on each periodical, explained Tom Pai, EPC Solutions’ project manager. When the periodical was picked up, the system detected that action and the software could then identify it was being read.
“Tagged journals will always be scanned when placed on display racks. Once someone picks them up, the system loses them on the scanned map,” said Pai.
The software then simply updates the status of the magazine as having been read, with a time stamp. “We will add one to the read rate count. This way, we can get a cumulative report for library to decide which journals to stop subscribing to,” Pai said.
Capturing Specific Read Data within Racks
Each magazine has a passive UHD RFID tag affixed to it, encoded with a unique ID that links to the details of the magazine in the web-based software. To enable the tag capture when magazines are on the shelf, EPC Solutions installed two readers and eight antennas for each rack. EPC Solutions manufactures its own hardware and provided its HL-CW300 4-port readers and its EPC antenna FlexAnt.
The company’s patented FlexAnt antenna is flexible and long enough to capture tag reads at short range, anywhere on the rack’s shelving, while being hidden behind the shelf’s décor, Pai explained.
The software, on a web-based server, leverages AI to identify and extrapolate from each read event. “This is all about human behavior,” said Pai, so the technology needed to be designed to detect and understand specific activities.
“Some people pick up the magazine and return to their seats to read, some pick up the magazine and stand in front of the bookshelf to read, some hold it higher, and some hold it lower,” Pai pointed out.
The software therefore was designed to capture the various movements and handling of magazines and over time identify what is taking place. That is possible with AI as well as Random Forest machine learning, technology with RSSI, timestamp and the antenna number for each tag read. “AI can use that information as well to filter out any stray reads,” said Pai.
Understanding Time Date and Behavior
The library’s goal is to learn not only specific magazines’ popularity, but types of magazines. That information could thereby influence additional subscriptions. The system tracks the reading cycle as well.
“It will help [the library] understand how interesting this magazine is,” based on how long readers continue to read it, said Pai.
The system will identify which times, for instance days of the month, in which each periodical is being read. For example, some magazines may be read almost as soon as they are received and put on display by the library. Based on how much the magazines are being read, the system can also make recommendations related to which magazines may need to be stocked in larger quantities.
The library intends to continue offering its physical magazines and journals as well as the digital ones.
Potential to Expand Across All Libraries
For libraries, Pai pointed out, having visibility means “better control of subscription budgets. Otherwise, you can only subscribe randomly with your eyes closed. It would be a waste of money to subscribe to magazines that no one wants to read.”
Since the project was initiated by the Digital Industry Administration, if the technology is successful, it has great potential to be promoted nationwide to other public libraries, said Pai.
That could mean other benefits such as analysis related to patron reading behavior in specific libraries. That information could help specific libraries better plan for subscriptions that are appropriates for their location.